Wikipedia talk:WikiProject Elements

From Wikipedia, the free encyclopedia
Jump to: navigation, search
(edit · watch · refresh · history · Old notices · RC for main articles · Article alerts)
WikiProject Elements (Rated Project-class)
WikiProject icon This page is supported by WikiProject Elements, which gives a central approach to the chemical elements and their isotopes on Wikipedia. Please participate by editing this article, or visit the project page for more details.
 Project  This page does not require a rating on the quality scale.
Article alerts

Requested moves

Featured article FA  A-Class article A   GA  Bplus-Class article Bplus  B-Class article B  C-Class article C  Start-Class article Start  Stub-Class article Stub  Featured list FL   List  Wikipedia Book Book  Category page Category  Disambiguation page Disambig   Draft   File   Portal   Project  Redirect page Redirect   Template   NA   ???  Total
24 7 71 4 33 92 88 31 1 178 131 262 5 0 118 1 16 3,377 230 0 2 4,671


Category Color Set review[edit]

For earlier discussions about categories & background color: see /archive_22
The periodic table using 11 colors for metallishness categories

The current set of Category legend colors we use in periodic tables at this wiki (enwiki) has some serious flaws, especially with regard to good webpage design. For example, background colors may have too little contrast with regular font color (like the red (alkaline metals) in group 1).

This topic is opened to review and improve the set. Starting point is the set as defined and used, labeled the 2015 category color set. This set is used throughout the enwiki consistently in all periodic table graphs (dozens).

Category is the name used on enwiki for these groupings (classes). There is no distinctive word for it (sometime "series" is used, but that can also refer to other classifications of elements). The definitions and compositions of the categories themselves is a scientific topic, and is not up for discussion here. There is no overall setup for this discussion. Section structure is our friend. -DePiep (talk) 12:40, 9 January 2016 (UTC)

Known issues

  • Contrast bg - font (including colroed fonts)
  • Distinction between category cellcolors (category border)
  • Contrast with background (eg yellow on white), esp in micro pt (no celltext)
  • Predicted category property (eg E119): no related color available
  • Add option ENGVAR (color/colour) in legends (not re Al, Cs etc).

Useful links[edit]

Note to all: added option to preview mobile view on your desktop. Recommended! -DePiep (talk) 10:19, 10 January 2016 (UTC)

Potential considerations in color selection[edit]

This is a brainstorm of potential principles or considerations. Others are welcome to add more. Ideally, items should be stated in such a way that people can vote them up or down. However, our goal at this point should be to collect as many different principles as possible, without taking the time to judge them or comment on them at this point. There is no need to sign individual suggestions -- adding an item to this list doesn't mean that you agree with it -- in fact, you may actually disagree with it. Consequently, there is no need to sign individual additions to this list.

Please do not comment on the merits of the items in this list at this point. Let us delay that discussion until we can all agree that this list is complete and includes all possible considerations - or at least all of them that we can think of at that point. YBG (talk) 05:20, 4 January 2016 (UTC)

  • Colors should be selected to provide adequate contrast with the font color
  • The palate should be selected to minimize changes from the current palate
  • The palate should be selected without regard to the current palate
  • The palate should provide adequate contrast between all pairs of colors
  • The palate should provide contrast between colors that appear adjacent in the PT
  • The palate should not over-emphasize any one part of the color spectrum
  • The palate should consider the mnemonics associations of colors where possible
  • The palate should not consider any cultural meanings of colors
  • The (non)metal color should be chromatically similar to the colors of the constituent categories
  • The (non)metal color should chromatically contrast with the colors of the constutuent categories
  • The (non)metal color should be totally independent of the colors of the constituent categories
  • The (non)metal color should not be considered in this discussion
  • The palate should ignore all existing associations, knowledge, current situation
  • The palate should consider the internal and external relations (of the categories being colored)
  • The palate should be selected using professional guidelines in coloring
  • Single step only: Let Sandbh decide from memory what his highschool wallsize periodic table colors were and use those. After that Sandbh will find so-called sourcing and convince all others that they can not read a graph (it worked with group 3 descriptions, recently).
LOL! That's a good 'un! Seriously, I can only remember monochrome or duo-tone tables in those days. The first fancy table I saw was the two-sided Sargent-Welch version that used to be bound into the Journal of Chemical Education, once a year. Sandbh (talk) 10:40, 5 January 2016 (UTC)
Don't let lack of sources stop you. You'll find a way to push it. -DePiep (talk) 18:29, 7 January 2016 (UTC)

R8R's color choice principles[edit]

While all of these were not a part of the initial design specification, these are the principles I have discovered while doing the work. Feel free to agree/disagree on each.

  • The TMs and the f-block categories can't be shades of the same color. They total for 68 cells of the 118 we have (fewer still if we exclude, for example, the gray ones, which are meant to stand out), and if they were one color, it would dominate the table, which is one of the reasons we consider a recoloring. One the other hand, it can't be too different as that would look not very pleasing aesthetically (so we can't have green TMs, red lanthanides, and brown actinides).
  • More obviously, the colors of Ln and An should be shades of one color, but these shades should differ significantly enough to make them impossible to mix up.
  • The hottest/warmest colors found in the table should be found in the p block, as that's the place with the most diverse and intense chemistry.
  • Yellow is a great choice for the metalloids. As the category is meant to be the intermediate of the p block metals and the nonmetals, the two should be completely different, with the color being between them, while being the in-between, should be independent from both PTMs and nonmetals (not too similar) and look like a natural delimiter. (So green-yellow-red is better than, for example, yellow-orange-red) This would allow using them for the metals-nonmetals table as well.
  • The color for the noble gases should be retained or a similar one should be found, as it suits the category logically and it differs the category from its neighbors, just as they are greatly differed by their chemistry.

--R8R (talk) 13:12, 5 January 2016 (UTC)

Continuing as per more work with colors:

  • The spectral coloring idea remains a good idea for me. This has been proven to me once again by Double sharp, who at one point declared the difference between the diatomic nonmetals and the noble gases was great enough to see the difference between neighboring F and Ne, but not great enough to easily determine which one hydrogen (quite far from both) was. The spectrum allows keeping similar colors next to each other and not far away. (Hydrogen, of course, remains a problem for this, but the colors have been worked since.)
  • I have considered the idea of simply dividing the 360 degrees into equal sectors on the hue wheel, but I decided against it. The major reason is, equidistant colors don’t necessarily look equally different. The problem is especially remarkable with shades of green. 150 degrees and 180 degrees are much more similar than, say, 30 degrees and 0 degrees. For this reason, equal hue distances have been dropped in favor for manual color setting.
  • White is not used. I have tried white and decided against it. The color is too bright and it is thus attracting too much attention, compared to the pale colors for the other categories. However, it did influence me in that I’ve picked a brighter shade of gray for “unknown”.
  • While generally, saturation (s, can normally vary from 0 to 1) and value (v, can normally vary from 0 to 1) are constant throughout the table, it is completely okay for looks to slightly vary them. While normally s=0.25/0.15 (regular/predicted) and v=1.00 for most colors, some colors may vary from these fixed values by no more than two percent. Such small differences do not ruin the color balance and may possibly improve it, especially given the contrasts that may be found in the hue wheel (for example, h=180 is much brighter than h=240).
  • This scheme has a color for characterized superactinides. It is established because the scheme is supposed to serve more than just a couple of years, and in a possibly not so distant future we may get early superactinides synthesized.--R8R (talk) 14:42, 31 January 2016 (UTC)
"The hottest/warmest colors ..." No, that is an undesired cultural meaning which does not hold. Also, other colors do not have a similar 'related' meaning which makes the selection unbalanced in this. -DePiep (talk) 18:13, 7 January 2016 (UTC)
Same for "suits the [NG] category logically". -DePiep (talk) 19:14, 7 January 2016 (UTC)
  • Expressed by HSV, current proposed colors differ by both Hue and Saturation (and even Value once). That explains why it looks like a chaotic carnival of colors coming in from every direction. Also, it is not web-professional design. -DePiep (talk) 19:14, 7 January 2016 (UTC)
If I call this "different color, same tone", some good examples are in this list and this. (However, for other reasons they are bad colors: should be gradients in both cases). -DePiep (talk) 19:51, 7 January 2016 (UTC)
A great tip, thanks.--R8R (talk) 03:01, 9 January 2016 (UTC)
So, find a HSV calculator, set S and V (eg 80, 255 for starters) and pick 11 Hues on the circle (256/11 steps ...).
Don't forget: first drop each and every cultural/habitual ties to colors. -DePiep (talk) 06:02, 9 January 2016 (UTC)
  • WP:ACCESS mentions (or so), that is about color for maps. Next level of quality! -DePiep (talk) 06:02, 9 January 2016 (UTC)

  • re: unchanged spectral sequnce you say. Will see what it looks like (I liked the pair-switches or ever triple-switches). Adding superactinides: bad idea good idea, but big trouble ahead (2 Feb). They are just predictions. For after 'a couple of years' we can not know what we need. More likely the category set (now 10+1) might change (merge groups 1 and 2; merge non-reactive nonmetals) quite more likely. Remember, we already agreed (did we?) that predicted categories should have the lighter shade. Most problematic issue with this is that 10 colors is stretching the limit already too much. Occupying colors for a prediction is reducing the overall effect. I can agree to pick colors on the hue-cirlce not evenly, because is easily shows that the distinctive hues are not evenly spread (some colors use a lot of degrees without changin enough). -DePiep (talk) 19:17, 31 January 2016 (UTC)
  • 2nd: OK, good idea, but it gave me a bad sleep. eka-An require their own color in the main set, but that implies we need eleven category colors (plus 'unknown'=grey=okey). That is a huge requirement (even colorbrewer stops a eight). Expect we must drop other wishes, think: coloring a font red/green to show SoM, can't afford. See where this goes. -DePiep (talk) 22:00, 2 February 2016 (UTC)
I understand your concerns regarding the eleventh category. The main reason why I want there to be an eleventh category for the superactinides actually is, I just want the extended table to be as good as possible, and I think having a color for the category makes it look better designed. Even the legend must be good. (But then, since we have a color for that, I want it to potentially work, and not only look good in the legend.) I originally started off with 12 categories (+eka-superactinides), but that they're gone, things are easier to manage. The main concern was, the colors for predicted AEMs and superactinides should look to some extent distinguishable. I hope I got it? However, while I (of course) want them to be easily seen as different, I don't see how I could do more given the main AM and AEM colors need to be easily different, and the main AM and NG colors need to be even more easily seen as different (given they are so far away from each other), and these differences are far more important. But again, I consider my current result to be okay, given they secondary role of those. I was able to keep the colors for fonts, except both had to become a bit darker and green had to turn into blue. (Darker is not necessarily bad: I never liked that #ff0000 we had for gases, and they colors must still be clear enough.)--R8R (talk) 07:51, 3 February 2016 (UTC)
Also, yes, of course, if we lose one color for nonmetals (or some other one), the scheme will need to be re-adjusted, and just dropping it won't do. But it's a different kind of change: a new supeactinide is a continuation of the current category set, a change of categories means a different category set. We can (and should) prepare for the former case within the current paradigm, and we'll have the paradigm shifted in the latter case.--R8R (talk) 08:00, 3 February 2016 (UTC)
I don't quite follow what you write on distinguishing. IMO, we want neighboring colors to be distinguishable more easily (including NG-AM being circular/Janet), but not those that are separated. On top of this, I found this more important to have bigger differences in the p-lock, because the categories are smaller and more random (fewer elements in each and tetris-like shapes).
Folding those two non-metals into one: would give us colorspace we could use indeed. OTOH: if we want a set that is widely usable, this 'spare' color could be used in other wikis and outside for those who still use the 'halogen' category (they need two colors there). Anyway, as long as there are two cat's, we should deliver two colros.
About eka-An: a new point. How come we have horizontal borders between those categories Ln/An/eka-An? are these periodic differences actaully category differences (metallic properties, ...)? Or is it just an old historic habit (because discovery-era differ)? If we group Ln+An together in one category, then eka-An is just an extension of that group. (So can have one color, be it regular or lighter for predicted elements). Time to ask in the other category section (on this page)?
Minor: for SoM, you left the green font and use a blue one., Of course, that is confusing because it is close to wikilink-blue... The new red looks nicer indeed. -DePiep (talk) 09:16, 3 February 2016 (UTC)
In a spectral table, there are no similar separated colors (except for NG and AM. I never considered Janet... in that case, no similar separated colors at all). But again, I think that if we should fulfill as many requirements as possible, and only if that proves impossible, drop some (say, you can't color a political map of the world with a different color for each country... there are too many). I think we must try to have all colors different enough, and drop the idea only if that proves impossible. A few days after the beginning, I didn't think I'd ever have 11 colors as different as they are now. Besides, hydrogen is a problem that demonstrates the problem well. Having all colors different enough will make a better impression on the reader (including me).
Re nonmetal colors: if we get all reactive nonmetals combined into a single category, I think we should have no color somebody else would use in mind. Besides, we must realize there are those who we can influence (entry-level audience) and those we can't (most professional chemists). There will always be those who will just follow us, there already are. They'll adopt our categorization as well. Besides, my initial attempts to see what this would be like showed it wouldn't look great. A good-looking table is the reason why there even was a call for a change, right?
I'll copy your question re lanthanide-actinide distinction to the #Combine lanthanides and actinides into one category section and will answer there.
Unfortunately, green won't work because of color contrast problems. Liquid is found both on the top of a greenish (TM) and a red (diatomic nonmetals) background. I had no other color left. (Of course, I tried making red not only red, but I decided the black-blue-red trio works best.)--R8R (talk) 10:01, 3 February 2016 (UTC)
OK, easy to clarify this: 1. we will not cater a color-set for PT's that apply a different categorisation scheme. 2. Probably the eka-Ans should be added. 3. Whenever the number of categories diminishes (like folding two into one reactive nonmetals cat), we gladly will rearrange the set (even if it was chosen & implemented a day before...). Because: one color less = better separation on the hue circle (greater distances possible). -DePiep (talk) 11:24, 4 February 2016 (UTC)
I agree on each point.--R8R (talk) 17:53, 4 February 2016 (UTC)
Now this tougher point. Given the extreme 11-cat-color primary requirement (plus grey for unknown), all other requirements take a second seat. If the state-of-matter (SoM) font coloring does not fit (e.g., because is has not enough contrast with any cat-11, or because blue looks too much like wikilink-blue): then we forbid these fontcolors (they must be black or linkblue)! We won't compromise the 11-plus-grey set. Even the simple 'predicted=lighter' principle might be impossible to reach. -DePiep (talk) 16:43, 5 February 2016 (UTC)
Yes, but only if we have no other choice. Actually, this looks like a job a computer could solve (or prove impossible to solve), but I'm going to be short on time during coming weeks.--R8R (talk) 06:56, 9 February 2016 (UTC)

────────────────────────────────────────────────────────────────────────────────────────────────────Not sure if this is the right place to put this comment, but I trust you'll excuse me if I missed finding the right place. Anyway, I recall that one of the things that is done to help with color blindness is to avoid using hue as the sole distinction, especially with certain problematic colors. This idea could be helpful to us if we used some other method of distinguishing 'known' from 'predicted'. This would free us up to use more than just hue in creating distinction. Some alternatives:

  • Completely avoid predicted categories (i.e., mark them all gray), but that is probably much too extreme.
  • Just color a portion of the cell (e.g., the background color for the atomic number) or to use a color swatch before and/or after the atomic number
  • Use the normal color but put a dagger or some other marker in the cell and have the legend include something like "† indicates predicted category" or the like.
  • Color the cell background white or gray but use the cell border color to indicate the prediction

I've listed all the ideas I could think of, even ones I don't think have much possibility, in hopes that it might help stimulate other out-of-the-box thinking. YBG (talk) 09:46, 12 February 2016 (UTC)

re color blindness: yes, that is a trick. Especially red and green look alike for a colorblind person, so better not use them both. re predicted categories: I proposed to put that topic in second priority, because we first must find eleven good colors for the main job. Note that the issue you mention also occurs in the existing categories (so far). So, if we have deducted 11 good colors for the categories, we can start about predictions & their added issues. (btw, colorbrewer is a good further reading on map coloring. Introduces three different tasks for coloring (sequential, diverging, qualitative). They do not use red and green together in any option set; but hey they don't have to solve the 11-color problem eather ;-). Today, our proposal separates red and green by the yellow (metalliods).) -DePiep (talk) 10:03, 12 February 2016 (UTC)
My point is to provide more flexibility for selecting 11 colors by completely removing the idea -- whether overt or just in the back of the mind -- that the colors must be selected in such a way that each one can have a lighter analog for the predicted colors. YBG (talk) 10:13, 12 February 2016 (UTC)
Agree, and for the same reason. -DePiep (talk) 10:55, 12 February 2016 (UTC)
Great! I presume that everyone is doing their best to take full advantage of all color parameters and not (even subconsciously) reserving some degree of freedom for potential use for the predicted dimension. YBG (talk) 07:16, 13 February 2016 (UTC)

DePiep's color principles[edit]

  • See User:DePiep/pt-2016 (in mobile view)
  • Current color set: DePiep-4: Swapped color-pairs (swapped the two greens etcs): contrast in p-block! Hues taken from the R8R-set; ~Sequential now all incl NG. Tweaked a bit (slightly away from R8R). Use white not grey for unk.
  • 21:23, 9 January 2016 (UTC)
Adjust limits for small font-size we might use
  • color blindness (esp red-green)
  • type of coloring (range?, quantity?, ...). consult Brewer
  • Map coloring design rules/guidelines (colorbrewer, ..)
  • Incorporate properties of Categories
  • Other requirements may put in requirements to solve (SoM, border, unk property, ...)
  • Cover scientific base well & correct
  • Use web-design rules & guidelines
  • Make it international: The Wikipedia PT Category Color Set! (that's enwiki, we know ;-) ). Our color set may be fit for usage outside of enwiki. That is: in other wikis (who may not even bother thinking about this), and in the Rest of the World. That requires professional considerations (like ColorBrewer does).
Document it, both by science and by webdesign
  • Be explicit on limits & dont's, eg re border-element issues (example: exact list of elements in metalloids is not fixed). So far, such dispute/uncertainty/grey area we do not show in here (yet?), and so should be excluded from this (in its documentation).
Traps to be avoided
  • no cultural meanings
  • no historical meanings (eg 2015 colors)
  • never too loud. Always apply: lighter = better. (When working on this, one might think categories are Very Important and so deserve strong colors. They are not important. Colors should not overshout eg PT structure).
  • Don't check one situation (PT). check variants
  • Use HSV not RGB color space. Know hue.
  • today: HSL (for lack of HSV calc templates)
  • Started with the same hue per category as set by R8R for now. Set S and L as some starting value.
  • Today 09:39, 9 January 2016 (UTC), had to use HSL instead of HSV. Requires different use (meaning of S,L)
  • change text 'unk chem props' into 'unk' - this is about the one property only
  • Later, try elevating to three-cat colors for metal-metalloid-nonmetal
Color rules
  • see R8R ideas
  • periods are cyclic (Janets left step), not just standard 1-32 order
  • predicted color to be a lighter shade
  • consider: unk=white not grey
  • consider: different colors (by shade) for non-text variants (like the micro PT: darker)
  • research: trend not checkered
  • moid as border color (see brewer)
Restraints (requirements)
  • Trend
  • 11 categories + unk
  • M'oids as a
  • 'Predicted' property (now: a ~lighter shade of cat color)
  • Distinguishable between them (neighbor borders, cell-to-legend)
  • Contrast rules w3c re font (dflt=black, blue=wikilink 2x)
  • re State of Matter (SoM; now uses colored fonts black, red, green, grey)
  • re Occurrance (now: in cell border; uses dashing/dotting lines + blue)
  • re Redlinks (now: not present; not considered)
  • re blocks (now: no interference)
  • guidelines
  • color-blindness tests
  • Works well with multiple size (cell size, font-size)
  • color perception (better distinguishing, lightness).
  • language & color ...
Degrees of freedom
  • No historical meaning of a color (& not 2015 colros)
  • No cultural meaning of a color
  • Use classification types
Later, related
  • Use derived set for 3-class metal-moid-nonmetal (+unk) cats

  • Investigate & explore the trend-coloring (flowing hues by rainbow sequence, not checkered as was 2015 and R8R today)
  • more checks re background.
  • consult WP:COLOR poeple / or mw
  • copy/maintain steal useful principles R8R pointed out
  • Look at other wikis. Do they use 11 cats? Any effects we could consider?
  • describe category properties (from science)
  • Create help template "Explain cat color"
  • Document it for i18n: build Help, /doc, Periodic table category, automate

-DePiep (talk) 09:39, 9 January 2016 (UTC)

DePiep color calculations (RGB and HSV/HSL)[edit]

Some early notes on color calculations. About HSV HSL color spaces (=3D definitions, as RGB is)

(Issue: we have no HSV-calculator templates yet, so I use the HSL color calculations everywhere; same principle).
1. Colors are fed to a webpage in this RGB form: background:#AB4FF7 (that's 3 pairs of hexes: 00-FF for Red, 00-FF for Green, 00-FF for Blue. Each 00-FF is 0-255 in decimal notation). But. This form is unhelpful to tweak colors ("lighter please" - cannot do that by intuitively number changing). So we turn to HSV/HSL
2. HSV colors: the Hue is the natural color (picked from the rainbow). See HELP:Using colors for bright hues. The Hue number is in a range 0-360 (when degrees), and circular. 0=360=full red.
S and V are the settings for brightness and lightness (of that same hue). Usually in % 0-100 each. Examples: the support header/bg colors for our main page are toned the same way (though with more extreme S,V numbers than pastel); this set (see its update changes).

This S,V workings is difficult to grasp for me to, I only play around with their settings ;-).

So I pick 11 hue numbers from the circle (~360/11 numeric steps initially, so hues 0, 33, 66, 99,...). AND I set a single fixed S,V pair for all (say 0.75, 0.85). This way, their tone is the same for all (the ~pastel-teinte today).
Per color, the H-S-V set is recalculated automatically into #RGB color number and fed to the templates ('HSV-to-RGB' in my sandbox Element-color-2016 list).
3. Next: playing around with the Hues and SV's. Mostly shifting Hue numbers in the legend 11.

So, when we end up in the details and fine-tuning, you can refer to the hue number (mentioned in my sandbox), the lightness (SV-setting), or write "between AE and AEM bad distinction". I can relate to the 'violet' names you use, but that may change over time etc. Sure I will spend more time on all comments on this page. But sure, we have an avelache of topic roiight now (I postpone SoM font coloring, font-size effects, tough colorblind-test, b/w print effect, ...). -DePiep (talk) 09:10, 11 January 2016 (UTC)

@DePiep: I have written the templates for conversion from HSV to RGB: {{HSVtoRGB.R}}, {{HSVtoRGB.G}}, {{HSVtoRGB.B}}, and {{HSVtoHEX}} for getting a hexadecimal 6-digit code for a color as expressed in the RGB color space. Feel free to use them if you need to.--R8R (talk) 08:07, 3 February 2016 (UTC)

The R8R sandbox[edit]

The task is more difficult than it seems to be. I have established a sandbox, and if anyone wants to use it, please feel free. The preview table, the page with the colors. Unfortunately, you'll have to wait some time (10 minutes or possibly even more, I think) for the colors to take effect. (But then there's also the draft. The server asks for an email address to send a download link to; I could've found a simpler thing, just didn't bother to spend the time.)--R8R (talk) 01:31, 5 January 2016 (UTC)

I have a new coloring scheme in my sandbox, as can be seen here. All labels should be readable; all categories should be distinguishable; no color should prevail over the others. Please take a look, leave a comment, etc.--R8R (talk) 02:12, 5 January 2016 (UTC)
The scheme looks OK but it relies on categorising the alkali and alkaline earth metals into one category. I don't have any major objections to doing that subject to agreeing a new category name and hearing from others. At the same time I would like YBG to continue his overall approach to the recolouring question. Sandbh (talk) 11:27, 5 January 2016 (UTC)
It does not; but I have made the alkali metal color a little brighter still. Is the difference between the AM color and the AEM color clearly seen now?--R8R (talk) 12:40, 5 January 2016 (UTC)
I have reworked a few colors and now everyone's welcome to review it. I think I'll leave the table where it is for now (unless there are comments signifying problems, of course). Didn't take too much effort, but I now think it looks much better than yesterday. I am offering this as a candidate for a future color scheme, granted a decision on that will be adopted only after we reached decisions of structural questions (e.g. procedure, nonmetals, etc.)--R8R (talk) 19:19, 5 January 2016 (UTC)
I'm looking at the updated table on my iPad and still can't tell the difference between the AMs and the AEMs. Sandbh (talk) 21:36, 5 January 2016 (UTC)
Hmmm. A great idea, it didn't occur to me colors may look differently on different devices. I took my iPad, and have further lightened the AM color so now it's pink, light but not pale yet. I am tempted to learn if the difference is clear now?--R8R (talk) 21:47, 5 January 2016 (UTC)
The pink for group 2 is better. Now the colour for group 2 is too close to the colour for the transition metals. Also can you update the colours in the legend bar below the big table? Sandbh (talk) 01:19, 6 January 2016 (UTC)

I find this gorgeous. Thank you, R8R! Finally I can see the difference between "unknown chemical properties" and "post-transition metals". Even if the colours are hard to distinguish for some people, they can hover over the cells to get the tooltips. I love how the softer shades take up the bulk of the table, making it easier on the eye than it is now. (Although maybe there are too many greens?)

Here's a list of colours that have been taken in your scheme:

  • Alkali metals: pink
  • Alkaline earth metals: violet
  • Lanthanides: spring green
  • Actinides: green
  • Transition metals: blue
  • Post-transition metals: chartreuse
  • Metalloids: yellow
  • Polyatomic nonmetals: orange
  • Diatomic nonmetals: red
  • Noble gases: cyan

Double sharp (talk) 04:28, 6 January 2016 (UTC)

Yesss, grey for PTM was bad. Good riddance. For unk's, it might be OK (still a shade to pick), or we can use white for unk. To be explored. -DePiep (talk) 10:58, 9 January 2016 (UTC)
This colour scheme is growing on me. I like the fact that the s-block metals are pink-violet shades; the transition, inner transition and post-transition metals are green-ish shades and the metalloids, poly- and di atomics are at the other end of the spectrum. Sandbh (talk) 11:12, 6 January 2016 (UTC)

Thank you! My initial fear was that the scheme had too many blue shades, from which I tried to move, and never thought green could be a problem. We have 39 green cells now (15 Ln + 15 An + 9 PTM), and 41 blue ones (35 TM + 6 NG). So I still don't think it's a problem, is it?

Right now, I want Sandbh to help me determine whether all colors in the regular PT are easily distinguishable (by the way, please give it a look now, I have a little reworked the colors for groups 1 and 2), and I can look for predicted colors after that. We practically need AM(p), AEM(p), San(p), San, Eka-San(p), TM(p), PTM(p), metalloid(p), diatomic(p), and NG(p). We already have the predicted color for the NGs, and my initial feeling is that the other colors should be easy enough to find; I already have two more shades of green for the superactinides, and if more green is undesirable, then there's also beige, which I have originally planned for the eka-San, but finding three shades of beige and/or brown is also doable. However, these predicted colors are secondary to our regular colors, so I'll finish them first.--R8R (talk) 13:49, 6 January 2016 (UTC)

"I want Sandbh to help me ..." - No. "Looks good to me" is not the way to determine good colors. "Looks good to Sandbh" even less - by what authority? And, given the non-cooperative habit Sandbh expesed last half year (we still have to repair the damage in Periodic table, Readers are struggling as we speak), I have no confidence that Sandbh can judge and seek consensus on this at all. In this, I claim "no consensus" for what is means: change not accepted. -DePiep (talk) 07:58, 9 January 2016 (UTC)
Sandbh had said some time before I posted this he had difficulties with color perceiving. I have checked by color against color blindness, they are okay for the red-green blind and not for the yellow-blue blind. I'll give it a try but I don't think we can fit everyone in this case, as much as I want. back at that moment it didn't occur to me there's software for this.--R8R (talk) 10:02, 9 January 2016 (UTC)
Good start then, but for final tuning & correctness we need those tools and guides. I guess over at WP:COLOR or at Wikimedia there are people who shout know a lot about this (esp color perception). I am also enthousiastic about I saw you did not border greens & reds :-). -DePiep (talk) 11:04, 9 January 2016 (UTC)
The colours work for me. Looking at the current scheme and your proposed scheme, the former appears to be more subdued yet the contrast in colours seems to be better; the latter is more "loud" yet some of the colours appear (to my eyes) to have less side-by-side contrast. This is not big deal however. What do others think? A side-by-side comparison would be helpful. Sandbh (talk) 10:12, 10 January 2016 (UTC)
re side-by-side: sure needed in near future. For now: I've transcluded R8R's to my sandbox, crude start. A single variant set takes about >=6 dedicated templates, so it is cumbersome.
re impressions: I do like the new set being "different colors, same tone" (irrespective of the cat-assignings).
I'm not happy with the current sets being distinctive (though I proposed the spectrum-order). Esp the p-block is bad in distinguishing the tetris groups (which are less elements than TM and Ln+An). That is what we get when we want a sequence: green-yellow-orange ... In this, the 2015 does much better :-( . Also, looks different when checking smaller PT's (navbox, micro; and in mobile view). Maybe need more hybrid order: contrasting in p-block, spectral outside? Sandboxing goes on. (and not happy -yet- with Ds proposal, showing, to make NGs outstanding). -DePiep (talk) 11:17, 10 January 2016 (UTC)
Very interested to try the latter idea, but I don't understand it. Could you please explain it in a couple of words?--R8R (talk) 22:46, 10 January 2016 (UTC)
(not a reply:) Headache or improvement? In the set, I swapped color pairs (the greens, reds, purples, blues; not MOIDs not NGs). More contrast next to MOIDs. See my sandbox. -DePiep (talk) 11:58, 10 January 2016 (UTC)
re R8R Gtrs. In the extreme we can either use the 'spectrum' over a period: colors move as the rainbow does: blue-green-yellow-red. OR we can make it 'checkered': opposing colors next to each other (eg red-green-orange-blue-green-yellow). While we prefer (well, I do) the spectrum for various reasons, it has this drawback: neigboring colors look like each other (light green-yellow-orange). That makes them hard to distinguish. This even more in the p-block, where the color groups are small and irregular('tetris'-like areas with few elements). The -important- metalloid diagonal does not stand out, it takes another look (and good eyes) to see its border. So I did this trick: within one pair of colors (pair orange+red; pair light green/dark green) I swapped the spectral order: the darker red now borders yellow. The darker green now borders the yellow. I think this improves the neighbour-distinction, esp in the p-block where it is needed. Shown now in my sandbox. -DePiep (talk)
Aha, I see. The spectrum is the best idea I can think of: Two similar colors are easier to distinguish when they are close to each other. I have known that, and I experienced it a number of times when building the current spectrum table: Hydrogen is quite apart from both poly- and diatomic nonmetals, and it wasn't easy to see which one it was grouped with. I have seen a proposal of YBG's double spectrum here; sounded great, but then I realized we may run into this problem again. Given this, I think if a color sequence is not spectrum-based, it must be completely independent from it: this possibly could give more freedom than the spectral layout does. I'll, however, watch your sandbox to see if the checked spectrum may work; for now, groups 2 and 18 are indeed too similar, and I want to add the PTMs and the lanthanides currently are as well.--R8R (talk) 23:33, 13 January 2016 (UTC)

Updating the legend is possible, but I yet don't know how. I'll figure it out soon, though.--R8R (talk) 14:10, 6 January 2016 (UTC)

No need for "soon". Consensus first. -DePiep (talk) 11:04, 9 January 2016 (UTC)
I just had a look on my iPad and it looks impressive. The legend works well. There is another approach for consideration which I'll post shortly. Sandbh (talk) 21:42, 6 January 2016 (UTC)
Every time I click from R8R's table to the current one I feel disappointed with the current colour scheme – I was satisfied with it for a long while, but now it just feels lame and inadequate. ^_^ This is really impressive work. (We just need predicted colours!) Double sharp (talk) 13:57, 7 January 2016 (UTC)
Thanks again. :) I have a draft for the predicted colors, here (these are the simplest thing that could be done, the same colors with distance from white approximately halved; how does the easy way work? how many colors, if any, don't work?). The colors for San and eka-San are not meant to be the showcase colors, I have a draft for them in a PSD file in my computer. But to work these two categories, we need to decide first whether we will use eka-superactinides at all, we have a section below for that discussion. I strongly advise we don't.--R8R (talk) 14:41, 7 January 2016 (UTC)
And then there'll be the question of whether we should create a regular and a separate predicted color for both San and eka-San. Now we only a single color for the regulars and the predicteds, from which I want to move (element, say, 121 may be a reality, plus the legend will look better). Doable either way; but first we need to decide whether eka-San are in or out.--R8R (talk) 14:48, 7 January 2016 (UTC)
Funny, every time I want to leave something for a while to fix some questions before that, I almost immediately do the thing. YBG, and whoever I may forget: pls take a look at the new colors and see if they are easily distinguishable AND please visit the discussions on combining diatomic nonmetals and polyatomic nonmetals into reactive nonmetals and on visual representation of the extended periodic table, possibly start new ones if you think there may be other questions worth rethinking. It would be very helpful and appreciated. Thank you.--R8R (talk) 17:37, 7 January 2016 (UTC)


Links: the main 18-column table, the extended table, the page with colors.

I have stuck to the general idea of spectrum coloring, improving it with small modifications, having further adjusted the colors over time, including the font colors for states of matter and the frame colors for occurrence, and I have resulted in a scheme that I consider to be a major improvement to the previous one. But now, of course, I need your comments to know if I'm right and if it's good; comments would be very welcome.

I have described the principles the scheme is built on above, where you can find both old and new principles. I've had the colors tested for contrast, for both the font/cell background contrast and the frame/cell background contrast.--R8R (talk) 07:29, 3 February 2016 (UTC)

Contrast checks (WP:Access)[edit]

About contrast between background and font color, and legibility (re in general & colorblindness)
  • Contrast check for red font (gases) fails. -DePiep (talk) 18:27, 7 January 2016 (UTC)
Fontcolors for gas, liquids, unk fail contrast check on all (existing today's proposed) backgrounds. Taken into account that it should suit smaller fonts too. -DePiep (talk) 19:04, 7 January 2016 (UTC)
DePiep, what does fail mean? As I understand, you rely on the contrast ratio; what is the lower limit do you use (as there are many)?--R8R (talk) 05:13, 9 January 2016 (UTC)
A dumb question, nevermind--R8R (talk) 05:34, 9 January 2016 (UTC)
All analysers use the W3C calculation + rules. (see WP:ACCESS). Don't know the numbers. In our case, I'd say we must be better (higher border number) because our fontsizes sometimes are smaller than the default in the w3c numbers. My COlorPicker also gives ratios for colorblindness (eg red + green are bad). -DePiep (talk) 05:58, 9 January 2016 (UTC)
──────────────────────────────────────────────────────────────────────────────────────────────────── Maybe this won't solve at all (colored font for legend). We could also add a rule: 'may not be used in smaller fontsize', or even 'not colored fonts at all' and find a new way to show this. Or, why not, leave SoM out completely. -DePiep (talk) 06:07, 9 January 2016 (UTC)

Metallicity as a trend[edit]

  • Since metallishness is a trend, the colors (hues) should be shifting left-to-right. Not checkered as it is now. -DePiep (talk) 18:27, 7 January 2016 (UTC)
    • Perhaps, but it's not a great trend. There are fairly important local differences, and actually I think the slight deviations from a pure spectral scheme help ease the load of similar colours. R8R's scheme has three greens, but it works for me because they're not adjacent. If they were adjacent, it wouldn't – at least for me. Double sharp (talk) 03:36, 8 January 2016 (UTC)
The local differences do not invalidate a trend in this. The main point is that in a period, any run-of-single-color is not split (all category-X elements are adjacent, there is never a category-Y element embedded). Another thing to note is that the periods themselves are a trend (putting the P in PT). While being a different trend (but unrelated?) using a scattered color set breaks this aspect visually. Especially since the colors are much more intrusive than the period. To the reader is suggested that the elemetns are thrown together like dice. That is bad. To add, I can suggest that the trend can 'pivot' around the metalloids (as R8R already has introduced). In general, we should not pick colors because "they work for me". And about the 3 greens, Ds: sure you will be able to see their diff. This works even more so when they are adjacent, as Ln-An shows. -DePiep (talk) 10:19, 8 January 2016 (UTC)
Found a new illustration helping me. My sandbox User:DePiep/pt-2016 now uses R8R's colors as is, but in a lighter shade (shades irrelevant in this). Seeing the legend, it undeniably suggests/states that those blues have something in common for being blue (but they just are TM, NG ?!). Same for the greens (Ln, An, PTM ?!). This is undesired. -DePiep (talk) 10:58, 9 January 2016 (UTC)
I question my own statement that metallicity (our categories) is a trend along the period. Since we do not have an article yet about the categories (comparing), I don't know if this is true. OTOH, supporting/suggesting a trend (by following the rainbow spectrum left to right), we do strengthen the Main Trend: chemical reactiveness (Period!). I could live with this effect. Also because it makes the PT easier not wilder (chaotic/checkered) in coloring. Topic to be revisited. -DePiep (talk) 09:49, 11 January 2016 (UTC)

With regard to metallishness/metallicity and the color spectrum, it seems to me there are two ways to spread the spectrum across the PT

  1. colors 0-1-2-3-4-5-6-7-8: In one series from left-to-right in the PT,
  2. colors 8-6-4-2-0-1-3-5-7: In two series from center-to-edge in the PT,

The second means that adjacent colors are further apart in the spectrum (except for 0-1). Have I explained this enough that you can understand the general idea? If so, does it merit further thought? YBG (talk) 04:43, 12 January 2016 (UTC)

Yes, I see. It is another more step away from straight spectral (restless). Minor, note that when circular (group sequence 17-18-1-2) it has 5-7-8-6. Will dive into this. -DePiep (talk) 10:21, 12 January 2016 (UTC)

Language and color recognition[edit]

This does depend on how your native language as a viewer breaks up colour terms, so it could make a difference on other WPs. In Chinese for example pink is treated as a variety of red and not a basic colour, so AM and diatomic nonmetals in R8R's scheme are going to be perceived as distinct shades of red rather than two separate colours. I don't think you can get around this for all languages because of those languages with very few colour terms, so if we are exporting this scheme to other Wikipedias, I think we have to admit defeat on this one and simply choose colours that can be easily distinguished – at least as different shades. It's not a vital thing like catering for colour blindnesses; we already have multiple shades of red in the current scheme, and nobody complains about it. Double sharp (talk) 13:06, 9 January 2016 (UTC)
Guess I don't understand this. What I tried to explain: the legend has two blues for TM, NG (R8R's, mine, ...). We assume they are distinct enough for the purpose. Al fine so far. But when I see the legend/PT, I see two blues and subconsciously conclude they are related, and less-related to the other colors. This blue=blue connection is independent of culture (language, education, habit). The same for green colors. Now when the colors jump in the period (as these examples do), that is an undesired effect. (I actually do see a 'bridge' over the nonmetals, a connection/association/relation. Also a 'bridge' across TMs for the greens. Then I go: what is the meaning of that bridge? - there is none we know).
If I get you right, you say that language of the viewer is in play. I don't understand. How would the name or language-association for two colors steer the eye? Would a Chinese not see the blues as two blues? -DePiep (talk) 13:19, 9 January 2016 (UTC)
More strongly than you'd think. Give this a read: short article inside--R8R (talk) 14:23, 9 January 2016 (UTC)
A Chinese may even see the greens as related to the blues, because 青 (qīng) can mean either colour, although since there are unambiguous words (蓝 lán for blue and 绿 for green) it may not be the case. It would be more likely with pink and red, since pink is considered in Chinese to be a shade of red, so they'd see a link between alkali metals and diatomic nonmetals. OTOH maybe R8R doesn't see the blues as two blues, but as two different colours, because IIRC in Russian light blue and dark blue are non-overlapping, different words (голубой and синий). But I'd have to ask him about that. Double sharp (talk) 14:52, 9 January 2016 (UTC)
In general, голубой and синий are two entirely different colors. In practice, personally, I am too used to the English blue and the German blau, so I usually think of "blue" as of some intermediate color, and even in Russian, if someone says the word синий, I may ask, "do you mean the синий синий or голубой?", if the color is particularly important for the story. But in general, yes, these are two completely different colors.--R8R (talk) 15:41, 9 January 2016 (UTC)
Wish you hadn't told me this... (looks like it's not just the words used like "dark" for all darks, but the cultural implication to only recognise the "darks" as such in upbringing. Impressive). -DePiep (talk) 11:30, 10 January 2016 (UTC)
──────────────────────────────────────────────────────────────────────────────────────────────────── So there is a language(/cultural) perception in colors. Now I see it is everywhere:
(LOL/side topics ahead): this person is perceived to have two different iris-colors (but is ithat true, or is it a cultural perception - after all, he defined a new culture himself). How did he see the PT?
Does this relate to this story? Currently, iu:white is not even one word!
Now that you mentioned this, I remember that my mother could recognise & name 50 different browns (in couture), I know only two (light and dark). -DePiep (talk) 09:40, 11 January 2016 (UTC)
  • I am happily surprised that this first throw by R8R is already gaining support. Of course, given what we come from (the 2015 colors) this could be an easy achievement, but I am also happy that the mold is broken to use "red is for alkali because fire" and "because I am used to a reddish there" -- almost completely broken. That said, the current set is not fit for mainspace because of blocking issues I mentioned throughout. -DePiep (talk) 10:25, 8 January 2016 (UTC)

I have read a number of guidelines earlier today, and I think it's been a great change with the sandbox I've made. The colors have calmed down and the color order is finally straight rainbow.

  • DePiep Now that you're in the game, I may leave it just there if you want me to. However, I'd love to mention one thing: the colors, while providing contrast with the text, can't be too pale. While (of course) intense colors are certainly not desired, too pale colors are difficult to distinguish on my first look from each other, even though I'm a person with no eyesight problems. This causes eyestrain, which is one of the reasons why there are web coloring guidelines. Good luck with finding a color contrast enough from the current diatomic red; I had to darken the current #ff0000 a little to reach even aa. I'm afraid the aaa is unreachable with the current bright red: the contrast between #ff0000 and #ffffff is 4.--R8R (talk) 13:17, 9 January 2016 (UTC)
My whole setup is that you'd keep running. I need my ideas tested & criticised, I'm fine with alternative sandboxes. As we have discovered here, dozens of aspects must be taken into account, so at least it is a cyclic development process. eg I plan to evaluate each of the remarks re your sandbox against my setup (and use/steal every improvement along the route).
What we do need is time (I am afraid we'd push an early version for the low hanging fruit, and then it stops). I only have begun reading about bordering colors & distinctive colors. And that number 11 makes this a huge challenge! All pro tools need evaluation. I want to invite webpage-designers from wikimedia.
Maybe you are interested in building the support to get this well-based and spread international: article Periodic table category should be blue, etc.
Now for today I'm off; bizzy in RL next week. -DePiep (talk) 13:52, 9 January 2016 (UTC)
PSs 1. I started with the extra pale colors, knowing that they only would get darker in the compromises. It's a trading trick ;-). 2. Your showing confidence in my approach is well appreciated over here -) . -DePiep (talk) 16:18, 9 January 2016 (UTC)

General comments on the R8R sandbox[edit]

The lanthanides and noble gases (on my iPad) now seem almost identical in colour; the transition metals and the post-transition metals are also noticeably close in their shading. Sandbh (talk) 21:32, 11 January 2016 (UTC)
@Sandbh: Yes, I think they're too similar in R8R's current version.
What's your opinion on DePiep's sandbox? Double sharp (talk) 07:03, 12 January 2016 (UTC)
I like R8R's version except I would make the PTMs grey using the colour of the elements with unknown properties and colour the latter white. I think the contrast would be better that way, In the table above the colours of groups 1 and 2 are too close as are the TM and PTM colours.
Pls follow my sandbox, some evolution & playing in there (also re Ln/NG and TM/PTM you mention). Grey in the main sequence is breaking the setup: from hue circle; grey is not a hue, undesired. Pls note which PT's you'd like to have in a compare-testpage (expect 2 or 3 color sets in there in any PT). Somehow the AM/AEM is ending up too dark again, as in 2015 ;-).

@Sandbh: I've been tinkering around with the colours. I think the trouble with purely going mathematically, equally spacing the hue values, is problematic as the categories of English are not equally spaced. I think yellow is a pretty clear category around 60°, but it seems like almost anything from 90°–150° can be taken for green, and thus you do not see a difference between TM and PTM even though it is bigger than between metalloids and TM, because English distinguishes the latter (yellow/green) but usually not the former (chartreuse/green). I've tried to exaggerate the differences in the pairs you mention, although it drives the noble gases into a wall. Right now I can see the difference (to me, group 2 is a bright magenta while group 18 is now a dim rose, that can be told apart from the nearby dull red), but I'm not sure if you can. Double sharp (talk) 14:38, 12 January 2016 (UTC)

Top table is better now. I can tell the difference between all the colours (there's not much in it between groups 2 and 18). Sandbh (talk) 22:15, 12 January 2016 (UTC)

The DePiep sandbox[edit]

  • User:DePiep/pt-2016 (mobile view) by DePiep (talk). Please do not edit; discuss here.·
  • Current color set: DePiep-4: after some playing: full spectral, (NG joining in), Hue ca 30 deg/step. Kept swaps-per-pair. Change: white for unk category, change text in legend to 'unknown' only. -DePiep (talk) 18:55, 10 January 2016 (UTC)

Is this about right(?):

Alkali metal Lavender blue
Alkaline earth Lilac
Lanthanide Aqua
Actinide Pale blue
Transition metal Light green
Post transition metal Dark green
Metalloid Yellow
Polyatomic nonmetal Red
Diatomic metal Orange
Noble gas Pink
Unknown White
--- Sandbh (talk) 09:54, 14 January 2016 (UTC)

  • Sandbh and others: what would a compare-color-sets testpage look like? Which PTs, legends, info, and how? -DePiep (talk) 09:26, 11 January 2016 (UTC)

Predicted category colors[edit]

Split into separate subsection

Now how does this work with predicted shades? Can we have a Fricke extended table? Double sharp (talk) 04:28, 6 January 2016 (UTC)

We practically need AM(p), AEM(p), San(p), San, Eka-San(p), TM(p), PTM(p), metalloid(p), diatomic(p), and NG(p). We already have the predicted color for the NGs, and my initial feeling is that the other colors should be easy enough to find; I already have two more shades of green for the superactinides, and if more green is undesirable, then there's also beige, which I have originally planned for the eka-San, but finding three shades of beige and/or brown is also doable. However, these predicted colors are secondary to our regular colors, so I'll finish them first.--R8R (talk) 13:49, 6 January 2016 (UTC)

  • When main set design is made (following pro guidelines), ei use hue-only choice, the 'predicted' category colors are easily and correctly derived in one systematic step: finding the one lighter shade (calculation) for all of them. Is what we did/tried in the current set, but the reds forced us to us into the strange dark red for E119). -DePiep (talk) 19:42, 7 January 2016 (UTC)

Redefine categories[edit]

Colors used elsewhere[edit]

 • Encyclopædia Britannica (1997-2010)
13 graphs, heavily using category colors. No metalloid category. Legend colors do not match PT colors?!

Big bold colours[edit]


For a fresh look at a different colouring approach see here.

So, a bright and optimistic spectrum-based approach? (Also, no lanthanides and actinides? How naughty. They ought to be sent to bed without any supper. ^_^)
I think the colours are too dark. They look fine because the lettering is in white; but we, using normal black lettering, are going to have problems. Additionally, even if a spectrum is used (like R8R's), I'd prefer it not to be strict; I like how R8R splits the greens (a fairly obvious light/dark pair in the f-block is fine, but then he uses blue for the transition metals to avoid too many colliding greens, going back to chartreuse for the post-transition metals). Furthermore, I would argue that the noble gases ought not to continue the trend from either the right or the left. As one goes further towards the edges of the periodic table, the trend is towards increasing reactivity, a trend that is abruptly cut off with the noble gases. Given that these trends are not uniform throughout the periodic table, I think R8R's deviations from a pure spectral scheme make his scheme better. Double sharp (talk) 08:10, 7 January 2016 (UTC)
Agree with your opening lines. Disagree on the spectrum thing: needs spectrum for being a trend. "I like the greens separated" is not a coorect way to handle this. Professional web design, esp in geographic maps & region properties, can help us better. -DePiep (talk) 10:30, 8 January 2016 (UTC)

Outside of chemistry[edit]

Good mapping examples from elsewhere:

  • Now that the US election circus is touring, there are great infographics around. The NYT is famous, and one can check this one, just for inspiration: fivethirtyeight.
Note that later in the process, the number of colors (candidates) will be lower, so better take a look asap. -DePiep (talk) 21:24, 11 February 2016 (UTC)

What about white?[edit]

Is there no use for a white colour category? White is a colour too. Sandbh 09:59, 10 January 2016 (UTC)

Only when useful. However, just like greys and black (and brown, for some reason I don't know yet), it is not a hue, not a point in the rainbow. As such, it would be an exceptional color in the spread we want. So far, I only thought of using white for the unk 'category'. Must also take care of merging-with-tablebackground effect. (lol: and then add a "unk (predicted)" category?).
btw, greys, including a very light shade, are used as support colors: table title, background, box borders, column/group stressers. White, as a sort of grey, is associated with that. -DePiep (talk) 10:08, 10 January 2016 (UTC)

Actually, I think there is a use for a colour that merges with the table background. For quite a while (four years from 2006 to 2010 IIRC), element 117 was not discovered (though everything else in the first 7 periods was), and so it had no border, being simply undiscovered. I would argue that in such a case (which could happen today if, say, element 120 was discovered before element 119), the cell concerned should appear, but it should receive a blank background, defined in the legend as "undiscovered". It shouldn't have a border for the same reason. (We'd have this today for elements 119 and 120 if we magically all converted to left-step tables.)

Incidentally, while we're on the subject of how to handle discoveries of future elements, I think that when E119 and E120 are discovered, we can simply take the old table and add them in their places below Fr and Ra (8s). The addition of the superactinides will royally mess up all the 18-column tables, so that will be our grand opportunity to turn 32-column once and for all (since reliable sources are not going to have a choice). At that point, I'm not sure how much to show. It seems odd to have only elements till 122 discovered, for example, and show 50 more blank-background elements. So maybe we would only show up to the highest number we know (filling in only gaps), or maybe we would show up to the end of the series (thus 155, then 164, then 172). But we can worry about it once the first superactinide is discovered. Double sharp (talk) 14:26, 10 January 2016 (UTC)

This is about category colors first now, the cell background. Not other properties (SoM is handled elsewhere). I already changed in my sandbox the legend text to simple "unknown" from earlier "unk chem properties" (plural even!). Here it means: "the category is unknown", nothing about other properties let alone discovery. SoM unknown is handled elsewhere.
Undiscovered elements (119+): don't see a issue wrt the categories. 119 or 120 will be unk category first, and so be have the bg as 118 in this. Or a predicted category - fine, we can deal with that too. Occurrence (border) and SoM will a\rfe available too. Indeed, the only thing missing is a note for "undiscovered"
Use white intentionally: I don't favor now. We should say 'don't know' just as clearly as the others. Thats different from saying (nothing).
Into Janet left steps: I don't mind, great. I am keeping an eye on that the colors work in all those variants. For example, later we'll have to check if group 1 can be next to 18 re the colors. This is part of the general philosophy of the PT (as Janet uses correctly). 2015 colors do too.
White category: Now I also sandboxing the white color for this cat. Just experimenting. Still the issue could be the merge with table bg (no cell square visible); Even fatal in the micro PT which has no text/symbol at all. Maybe a grey would be better.
Together, these are 'constraints' we try to fulfill with the new set (each one is tying us further ..). If some can not be achieved, we'll have to compromise.
White for "category unknown" tested in sandbox. But invisible cat color would be confusing imo. Might need differentiation from general whitish background. Unknown elements 119+ have no specific notation for being undiscovered (same in 2015). -DePiep (talk) 19:19, 10 January 2016 (UTC)

DS ideas[edit]

One thing that the interruption of the spectral sequence creates is that we don't have a large patch of the table (TMs plus Ln/An below and PTMs after) the same colour. If we are going to match a spectral sequence, then

I think the argument for a spectral sequence is pretty convincing, and matches electronegativity, first ionisation energy and electron affinity (as well as reactivity!) pretty well (with a few bumps and dips that we can probably smooth out). The slight problem is that the huge dip in reactivity at the noble gases (and the unique combination of high ionisation energy and high electron affinity that leads to it) is perhaps compelling enough to break the trend just for them.

So if I were to edit R8R's scheme to be more in accordance with the spectrum, I'd have a spectrum running from the alkali metals to the halogens, with the noble gases abruptly muted in colour, and then the trend can start again in the next row. The trend is about chemical properties, so it makes sense to mute the colours for elements with limited chemistry at standard conditions (the noble gases).

So I'd choose something like (forgive me for not making my own scheme, since I can't really choose colours well):

  • alkali metals: magenta
  • alkaline earth metals: violet
  • lanthanides: azure
  • actinides: blue
  • transition metals: green (preferably a bluish green to distinguish it clearly from the right)
  • post-transition metals: chartreuse (greenish yellow)
  • metalloids: (pure) yellow
  • polyatomic nonmetals: orange
  • diatomic nonmetals: red
  • noble gases: a light cyan, close to white, just as present (abruptly contrasting with the preceding red)
  • unknown chemical properties: the current grey

This puts the cool colours on the metallic side, and the warm colours on the nonmetallic side. Since yellow is clearly differentiated from green and red, it stands as a reasonable midpoint. The sequence can easily wrap around from diatomic nonmetals (with the halogens as a subset) to alkali metals, with the noble gases standing in between these infamously reactive elements as a bastion of calm. Double sharp (talk) 13:06, 9 January 2016 (UTC)

This does run into the "huge patch of the same colour" problem for those languages that do not distinguish blue and green as different colours (just different shades), though. But R8R's current scheme already falls into that, and since they are clearly distinct shades still, I think we can live with it.
What do you know, I finish composing this post and check R8R's sandbox, and he's edited it to be almost exactly the scheme I outlined above. Nice! (Though I think the polyatomic and diatomic nonmetal shades are too similar; I'll try tweaking the latter a little, to push it a little further towards the reddish end of the spectrum.) Double sharp (talk) 13:09, 9 January 2016 (UTC)
Oh, right, that makes it collide with the red colour for gases (although we could change that if we wanted to). What I think could be improved about his current version is that TM/PTM and poly-/diatomic nonmetals are a little too close, and that the contrast could be exaggerated a little by choosing shades of the same colours (green/chartreuse; orange/red) that are further apart on the spectrum. Double sharp (talk) 13:12, 9 January 2016 (UTC)

With regard to the next levels in the g- and f-blocks: since actinides are darker than lanthanides, superactinides should be darker still. I think, like R8R, that we should seriously consider eliminating period 10 from the extended table, since the idea of element 184 is treated with caveats by Fricke himself, and does not appear in recent studies (e.g. Pyykkö), and it appears that the second island of relative stability would probably appear earlier, at Z ~ 164 instead. We're not really sure what happens beyond Z ~ 172 with weird things happening with the nucleus. Double sharp (talk) 13:24, 9 January 2016 (UTC)

I definitely gonna chew on all of this (but maybe not coming week). Likely will create the set in my sandbox. Whatever happens, spectral sequnce has its dangers (distinction of neighbors), so it must be proven right.
As for the M'OIDS=yellow can I agree (irrespective of the neighbor colors even), also because we can only use one shade of yellow (it is too light for a second distinguishable yellow). That can be the full yellow then. More later. -DePiep (talk) 16:13, 9 January 2016 (UTC)
Or maybe, since the superactinides are going to be in their predicted shade (and hence on the lighter end), we should perhaps have lanthanides blue and actinides azure, so that the superactinides can be lighter still. Double sharp (talk) 14:18, 10 January 2016 (UTC)
Played around with this. So far, 1. I did engage the NG in the spectral sequence. Having them standing out looked a bit unneeded to me. (and after all, they are in that place #18 rightly by sequence). 2. assuming the super-An will be "predicted" all so have a lighter An/Ln color. Not a main PT color. 3. In the 360 deg color circle for 11 colors, I made steps op 30 deg (Hue numbers 0, 30, ...), using some extra space for the separation of: NG, s-, f-block (ie colors wider from each other there). 4. Did do the pair-swaps again: eg in purple pair, blue pair, green pair, red pair. Not Moids, not NG. 5. And trying white for unk-category (was grey). -DePiep (talk) 19:27, 10 January 2016 (UTC)
re ""huge patch of the same colour" problem" (think TM): I don't think this is a problem any more. Now that we use "different color/hue, same tone=pastel-thing" (R8R adopted too), color are not 'attacking' any more (as some 1st try colors were: flashing!). To me, the whole table now looks almost equally soft. I see no need to avoid a color for a big category, any more. -DePiep (talk) 19:34, 10 January 2016 (UTC)

Hey, I like that noble gas colour. It follows the spectral sequence, but it's not actually a spectral colour. Nice touch. ^_^

Current colour scheme:

  • alkali metals: blue (I don't see it as purple)
  • alkaline earth metals: violet
  • lanthanides: cyan
  • actinides: azure
  • transition metals: chartreuse (on the greenish side)
  • post-transition metals: green
  • metalloids: yellow
  • polyatomic nonmetals: red
  • diatomic nonmetals: orange
  • noble gases: magenta

Could we see the superactinide colour in a mockup? It won't be a basic colour, but should probably continue the trend from Ln to An to San. The white looks good. I don't think the colours are attacking each other anymore, although we should probably test this (especially for colour-blind viewers – I accept that 11 shades for them is going to be difficult, but at least neighbouring shades should look appreciably different). Double sharp (talk) 08:08, 11 January 2016 (UTC)

All sounds good. quick Re: Neighbor-distinction is a main requirement indeed (so I'm very happy with the pair-swap eg those next to the yellow Moid for this :-) ). Wrote about color calculations. Super-An color can be done, but is one of the side-issues (like contrast with SoM fontcolor etc). Choice of priority & time-spending: I think the main and defining and core problematic issue is the Legend 11; other topics should follow or require a fine tuning. A circular design process, while learning. A more elaborate sandbox to come (compare sets, automated predicted colors, ...). -DePiep (talk) 09:20, 11 January 2016 (UTC)
Is there a possibility to move the PTM hue out to 135°? It would make the gap between 90° (TM) and 180° (Ln) smaller and also make it more distinct from the chartreuse transition metals to the side. This is the one pair I think may be a hair too close. Double sharp (talk) 12:23, 11 January 2016 (UTC)
OK, I tried that. Double sharp (talk) 12:28, 11 January 2016 (UTC)
Ds, is there some numeric hue space somewhere to pull groups 12 and 13 more apart? However, this has low priority, the other distinctions are more important (and better be kept). -DePiep (talk) 08:21, 13 January 2016 (UTC)
If we do that then either the TMs crash into the metalloids or the PTMs crash into the lanthanides. (I think it ought to be groups 11 and 12 instead, but we'll save that for another time.) Double sharp (talk) 16:07, 16 January 2016 (UTC)

Happy New Year with IUPAC recognition of elements 113, 115, 117, and 118![edit] (it's not 1 Jan yet for me, but it's already night). Double sharp (talk) 13:58, 31 December 2015 (UTC)

(RIKEN got 113; Dubna/Livermore got the others.) Double sharp (talk) 14:05, 31 December 2015 (UTC)
(Oh, and here is the 2013 article that criticised the IUPAC recommendation that -ium should be the ending for all elements, even those in group 17 or 18, which is why IUPAC proposes to amend it. I'm not too happy about naming elements based on chemical properties that are not even known, but I can't and won't argue with whatever they decide.) Double sharp (talk) 14:08, 31 December 2015 (UTC)
A marvelous quotation that should find an honored place somewhere in our series of articles:
YBG (talk) 06:20, 1 January 2016 (UTC)
I remember Scerri said the same some years ago. Too early ;-) ? -DePiep (talk) 16:57, 1 January 2016 (UTC)
  • ... and about the undiscovered elements. A bit strange that we don't have to change our graphs and infoboxes now that those elements are formally "discovered". I mean, element 119 does not state "not discovered" (except in text). Of course most of their properties are marked "predicted" OK, but the element it self not. -DePiep (talk) 16:57, 1 January 2016 (UTC)
    Good one. On the top of my head I can suggest simply add to History, "Discovery date: n/a." Possibly an even better idea would be to state the element is not yet discovered in the very header of the template, or right after the pronunciation. ("General properties," come to think of that, could use a renaming to more closely reflect it's just about the name.)--R8R (talk) 02:22, 5 January 2016 (UTC)
Or state something like "This element is predicted only", for the whole? -DePiep (talk) 12:38, 7 January 2016 (UTC)
I think a great idea would be to add a small "(predicted)" to the header of the infobox. So instead of
Ununennium, 119Uue
Ununennium, 119Uue
Thanks for the idea. I've had smth. like that in mind, but couldn't formulate.--R8R (talk) 12:51, 7 January 2016 (UTC)
But "Element 119" is not 'predicted' like a property, right? It just not created.
btw, we in the periodic tables we could just add brackets like (Uut) and (113). -DePiep (talk) 08:21, 8 January 2016 (UTC)
the name is not predicted, but the whole concept of element 119 is.--R8R (talk) 15:19, 9 January 2016 (UTC)

News from Eric Scerri

Great. Interestingly, he does not mention the third step in producing the Left Step PT: shift the s-block up one period (row). And exactly that is why elements 119, 120 appear in the seventh period - the point he makes. Maybe he should have consulted enwiki ;-).
Second: Separate topic, moved to Talk:Periodic table -DePiep (talk) 08:33, 8 January 2016 (UTC)
  • Happy extension with these four elements! (I forgot to celebrate in time ...). Four in one day!! -DePiep (talk) 19:45, 5 February 2016 (UTC)

Rethinking the list of our categories[edit]

A major visual change is coming for our PT. As such, it is a great time to either state the current categories are okay or change the set of categories.

Two possible issues coming to my head are a) combining group 1 and 2 into a single category and b) combining di- and polyatomic metals together. (Although any other issues are also worth a discussion, if there is one I haven't mentioned.--R8R (talk) 11:55, 5 January 2016 (UTC)

re "it is a great time to either state the current categories are okay or change the set of categories.": Redefining the (enwiki) set of categories must be kept a totally different topic from any presentation changes. They are independent topics. The graphics are e to represent the scientific facts, and nowhere the other way around. Understand that this un-tying gives a huge degree of freedom for both topics!
Having said this, any (scientific) redefining etc. can be developed now or in any other moment. The Graphic department should cover any scientifically well-bases outcome. Would there be a reduction of categories? Fine, we'll use or define one color less. Would a category be added? OK too, we'll add a color, systematically. Enjoy this freedom. -DePiep (talk) 10:01, 6 January 2016 (UTC)
Yes. I see the freedom. I perfectly realize graphics just represents science and should not affect it. The discussion would be equally valid any time, now or later; I chose now. Think of the phrase "it is a great time" as an introduction into the proposal of rethinking. There would be a different introduction if I didn't have this one, but I had it.--R8R (talk) 21:57, 6 January 2016 (UTC)
Indeed, can go simultaneously. It's just, no interference should occur in the development processes. -DePiep (talk) 11:31, 9 January 2016 (UTC)

Combine alkali metals and alkali earth metals into s-block metals[edit]

I'm not convinced about this given what you say below: the combined name is less well-known than the individual group names. Furthermore, if one were to write an article about the s-block metals, it would immediately break apart into two parts: one about group 1 and one about group 2. Double sharp (talk) 03:59, 6 January 2016 (UTC)
Neither am I, but since we revise our PT, I find it is best to list all possible changes and not just the ones I would easily support.--R8R (talk) 22:03, 6 January 2016 (UTC)

Combine diatomic nonmetals and polyatomic nonmetals into reactive nonmetals[edit]

I believe this would be a logical step. The set of the eleven reactive nonmetals is very diverse, that is true. However, I believe there is no use in breaking it in two (or more), because there is hardly a line all eleven could fall into either side of it. The current divide isn't one as well; one example showing that is iodine. It does fit into the C-P-Se-I diagonal trend; moreover, it is quite similar chemically to sulfur (for an element from a different group), and there is hardly a reason to keep the two apart.

Science aside, it is an even more questionable decision. One might ask, what's so different between C and Si, for example, and how does that differ from this case. The difference is in that "metalloid" is a long-established concept, and it is kept in out PT as such. "Diatomic nonmetal" and "Polyatomic nonmetal," being existing categories, are not nearly as often mentioned as a part of a standalone categorization; more often (and more obviously) they may be used as a part of a one-criterion categorization, with the criterion being structure of the substance formed by the element. All of our categories are well-known except for these two, which are "some nonmetal" and "some other nonmetal" (of other similar cases, the TMs and the PTMs clearly won't be combined, and the same is true for the TMs and the s-block metals. The s-block metals themselves are a different issue, but similarly, I only mentioned it for the sake of objectiveness, I think they should similarly be kept apart, since the individual names are better known than the collective name, and we are targeted at a wide audience), the the differing criterion being particular and not general. As it does not provide an unquestionable partition, we will be better off dropping it.--R8R (talk) 12:27, 5 January 2016 (UTC)

How about just hydrogen split from everything else (typical nonmetals)? Everyone treats H on its own and its chemistry (being the chemistry of a particle) is kind of sui generis. There is indeed a shift in the nonmetals from metalloid-ishness to pure nonmetallic behaviour but one can argue about where exactly it is, e.g. S and I, so that a fuzzy distinction may not be the best thing to show. Double sharp (talk) 03:59, 6 January 2016 (UTC)
I think it is not very accurate to say chemistry of H is "chemistry of a particle": hygrogen also has different isotopes, and essentially, it's the same chemistry. Nobody could identify heavy water as a substance completely different than water; the difference is only great for a difference between two isotopes, and does not approach a difference between the most similar elements. That means, chemistry of hydrogen is a chemistry of an element. (However, I can't yet say if I would support or oppose the idea of setting it aside.)--R8R (talk) 01:27, 7 January 2016 (UTC)
I think what originally convinced me though about this option (Sandbh's "Option 10" IIRC) was that the polyatomic/diatomic divide was a rather good predictor of metallicity, only falling down slightly for iodine. Then again, iodine is the heaviest halogen, so this should still not come as a surprise, whereas sulfur's incipient metallicity would come as a surprise if we focused it wholly on how close they are to the diagonal stripe of metalloids. It also falls down a little for hydrogen, which has a high ionisation energy (because the 1s electron isn't shielded) and whose negative oxidation state is unstable (because its single proton cannot fully stabilise the electron-electron repulsion forces in H), again because it is at the extremes of its category. Maybe we should not worry so much about extremes, which are going to be slightly odd no matter how you slice them (e.g. beryllium in group 2, or group 3 in the transition metals, or hydrogen and iodine in the diatomic nonmetals). Whenever you have a category, you will have trends in it. Not everything will be homogeneous (how boring it would be if every halogen behaved the same way). Though come to think of it this is in fact an argument to have fewer categories when possible. Double sharp (talk) 04:10, 6 January 2016 (UTC)
I think you just basically my thinking behind the idea of keeping the nonmetals together. I could only add how a break within the nonmetals is not needed: while metals are a huge group and thus very diverse, there are clear signs of some groupings: the transition metals, for example, are obviously different than the main group metals. The nonmetals form a small diverse group without a clear division within it, and I believe we shouldn't create one in such a situation.--R8R (talk) 21:57, 6 January 2016 (UTC)
Intermediate points (my 2c):
For introduction of those two categories: /Archive 15, option 10 (August 2013. Between other issues like At halogen-or-metalloid). While I am At this: of course, the word "halogen" should not be reintroduced in the categories. As is not proposed, so OK.
For me as a layman, these two cats are both correct and far-fetched (iow, does not mean much for The Reader). OTOH, at this first/global encounter of a PT level, the 'nonmetal' word & class is enormously strong. It may be the first property/word a pupil will remember after seeing our PT, and even be able to locate later in a wordless PT (more than group and period, I guess).
As with other details, the two categories could be/should be elaborated in an article (Nonmetal).
I like the name 'reactive nonmetals' (exclude NGs!), from the start.
-DePiep (talk) 11:48, 11 January 2016 (UTC)
I'd like to read the opinion of Sandbh on this. -DePiep (talk) 10:41, 12 January 2016 (UTC)
Sure. Leave it with me for a while so I can review the old history. Sandbh (talk) 10:49, 12 January 2016 (UTC)

(arbitrary break)[edit]

R8R, I see you are having a good Russian winter. There was a picture in our weekend paper of two icebreakers moving along the Moskva River, with the Kremlin in the background, during snowfall in Moscow. Apparently temperatures up there fell to as low as –20° C in the city. I hope you are rugged up.

This proposal would not result in better categorisation scheme. It would reduce the sense of wonderment associated with the current arrangement and dimish knowledge and understanding of the nonmetals.

The distinction between polyatomic and diatomic metals is easy to grasp, factually based and provides engaging insights. Briefly, the 'poly-' (Greek ‘’polys’’ = "many, much") refers to their multi-atomic molecular structures, their many allotropes, and their tendency to catenate or form compounds with multiple homoatomic links. The polyatomic nonmetals are thus 'poly-like' in at least three ways. On the sulfur and iodine question, the ability to catentate easily distinguishes the versatile chemistry of sulfur from that of iodine. Other differences in the properties of the polyatomic and diatomic nonmetals are set out in the nonmetal article.

The existence of diatomic and polyatomic nonmetals arises out of the interaction of atomic and electronic properties. Diatomic nonmetals form diatomic molecules due to either needing just one electron to attain a noble gas configuration or because they are small enough (N, O) to be able to form triple or double bonds to attain noble gas configurations. Conventional wisdom is that triple bonding is the limit for main-group elements. Since C needs four electrons to complete its octet, but is still a relatively small atom, it gets around this problem by forming three single sigma bonds and one delocalised pi bond (pi bonding being more characteristic of small atoms), resulting in graphite. The larger size of the remaining non-noble nonmetals weakens their capacity to form multiple bonds, via pi bonding, and they instead form polyatomic structures, featuring two or more single bonds, in order to achieve completed octets. So, the distinction between polyatomic and diatomic nonmetals, as well as being simple, is more fundamental than artificial.

More broadly, the taxonomic thread that runs through the three nonmetal categories is beautifully anchored in, and echoed across the remainder of the periodic table. Specifically, from left to right across an 18-column periodic table, as metallic character decreases, nonmetals adopt structures that show a gradual reduction in the numbers of nearest neighbours—three or two for the polyatomic nonmetals, through one for the diatomic nonmetals, to zero for the monatomic noble gases. A similar pattern occurs more generally, at the level of the entire periodic table, in comparing metals and nonmetals. There is a transition from metallic bonding among the metals on the left of the table through to covalent or Van der Waals (electrostatic) bonding among the nonmetals on the right of the table. Metallic bonding tends to involve close-packed centrosymmetric structures with a high number of nearest neighbours. Post-transition metals and metalloids, sandwiched between the true metals and the nonmetals, tend to have more complex structures with an intermediate number of nearest neighbours. Nonmetallic bonding, towards the right of the table, features open-packed directional (or disordered) structures with fewer or zero nearest neighbours. As noted, this steady reduction in the number of nearest neighbours, as metallic character decreases and nonmetallic character increases, is mirrored among the nonmetals, the structures of which gradually change from polyatomic, to diatomic, to monatomic.

As is the case with the major categories of metals, metalloids and nonmetals, there is some variation and overlapping of properties within and across each category of nonmetal. Among the polyatomic nonmetals, carbon, phosphorus and selenium—which border the metalloids—begin to show some metallic character. Sulfur (which borders the diatomic nonmetals), is the least metallic of the polyatomic nonmetals but even here shows some discernible metal-like character (discussed below). Of the diatomic nonmetals, iodine is the most metallic. Its number of nearest neighbours is sometimes described as 1+2 hence it is almost a polyatomic nonmetal. Within the iodine molecule, significant electronic interactions occur with the two next nearest neighbours of each atom, and these interactions give rise, in bulk iodine, to a shiny appearance and semiconducting properties. Of the monatomic nonmetals, radon is the most metallic and begins to show some cationic behaviour, which is unusual for a nonmetal.

That the terms "polyatomic" and "diatomic" will sound far fetched for the general reader is natural. This will be the case for all of our categories, with the exception of the "metal" super-category name. "Alkaline earth metal?" "Lanthanide?" "Metalloid"? What are these? I don't think this is necessarily an issue—it comes with the territory of classification science—as long as the terms are explained in more general language e.g. in the lede of each relevant article.

The html periodic table (HPT) does a fine job of categorising the nonmetals. In the literature the nonmetals are commonly explored in their vertical groups, which results in five or six "categories" depending on how boron is treated (either as a metalloid or a nonmetal). This is too many to be practical for colour category purposes and it overlooks cross-cutting patterns. Our HPT incorporates both features. It includes the named vertical groups, and the three nonmetal colour categories. Two of the latter traverse vertical groups; all three follow tangible lines of demarcation. I think the result is a "just so" mapping, with something in it for everyone: the general reader, the knowledgeable reader, and the expert reader.

In conclusion, I’m not seeing the advantages of merging polyatomic and diatomic nonmetals into a single category. Sandbh (talk) 06:37, 19 January 2016 (UTC)

Thank you for your concern. But have no worries: –20° is fine when you're prepared to such a temperature. A sudden fall of temperature (for example, from 10° to 0°) can be much worse than that.
And thank you for your detailed answer. It does a great job in providing understanding of what lies beyond the idea. And now that I've read it, I can finally formulate my main concern with the current scheme. The 11 elements in question do indeed show a great variety of their properties. But there is no universal division of these elements everyone would easily agree on. I'm certainly not advocating for such a rainbow in the upper right corner of the PT (or even its reduced state, with only the halogens standing out, which we once had). However, allotropy alone (and we only show that in the name) cannot be enough: this is not a proper overall (I can't stress this word enough) categorization; similarly, you wouldn't expect to see "BCC metals" in such a categorization.
But there are other differences between the two sets of the elements. I want to specify we are bound by Wiki terms, spirit, rules, etc. For that reason, I am not searching---and I'm calling for you not to---for the best table we can make; I am searching for the best table that would best fit the Wiki rules (in particular, WP:OR and/or WP:V). All categories from alkali metal to metalloid can be easily recognized as good for our purposes. Of what remains, the noble gases are commonly opposed to the rest of nonmetals in their nonreactivity, and are thus best left apart. At this point, we have the 11 elements in question. There is no divide that would fit them and be commonly found in the sources: even slicing p block into columns won't work, since a great part of it is already colored among the PTMs or the metalloids. Therefore, I believe it is best for our purposes to leave them together. One questionable category is certainly better than two. I pick a name of non-zero meaning and occurrence for that.
Separating four elements from a set of eleven, with no exact border between the four and the seven, which would introduce another uncertainty (sulfur/iodine), is somewhat questionable by my standards; but the worst of all is that we can't draw analogies with the other uncertainties. These ("are there 14 or 15 lanthanides?" etc.) are found throughout the sources; we have no choice but to pick one version, given the otherwise overwhelming recognizability of these categories. We now add one with an editorial decision to divide nonmetals by allotropy rather than anything. I do understand your reasoning, but the problem is, the divide has not been recognized by most other authors to be strong enough to be the divide throughout nonmetals, and thus I believe that for Wiki purposes these two problems (one more uncertainty and few RSes) are crucial.
Long story short: the divide may be a good idea, but it does not suit Wiki.--R8R (talk) 15:38, 26 January 2016 (UTC)

I'm still thinking about this interesting and difficult topic. In the meantime, I have some questions about your comments.

  • "But there is no universal division of these elements everyone would easily agree on." —Would it be fair to say that the nearest there is to a universal division would be "other nonmetals" (or "non-metals"); halogens; and noble gases?
  • "However, allotropy alone (and we only show that in the name)…"—Is anything more shown in "transition" metal? Or "lanthanide"?
  • "Separating four elements from a set of eleven, with no exact border between the four and the seven…" Could you elaborate what you mean by "no exact border"? Does sulfur's polyatomic structure and ability to catentate easily, for example, not sufficiently distinguish it and its versatile chemistry from that of (diatomic) iodine? Sandbh (talk) 11:31, 28 January 2016 (UTC)
  • Re other divisions: I'm not sure. I was overly confident when writing the original reply, but I have to admit I can't prove it. I'm leaning towards no, however, because in my experience, the group-by-group division is far more common (so technically yes, if it is left aside, but the gap is too wide). I'd love to do a quantitative research, if there was a way for me to figure how to perform one in a reasonable period of time.
  • "Is anything more shown..." -- both yes and no. Technically, of course, no; but in the chemical nomenclature, "transition metal" is a very basic term that denotes a set of elements that all match a single property (except the exact definition is a topic of discussion; but generally, yes. The term is so basic it cannot be more advanced than "poly-/di-/monatomic"; but unlike the latter, it has been extensively used for categorization purposes and is one of the most basic terms for that. Same for "lanthanide."
  • Of course. Both polyatomic structure and ability to catenate easily, which you mention, relate to allotropy; themselves, they are not enough in my opinion. yet on the hand, they are two elements with similar basic properties: very close electronegativities (2.58 vs. 2.66); ionization energies (compare I against S, P, and Se, and then Br: Molar ionization energies of the elements); both elements have a certain degree of metallicity, perhaps even more pronounced for iodine than for sulfur. One stronger difference is that a second electron allows for a much richer chemistry for sulfur in practical sense (sulfuric acid is well-known even for non-chemists, and a sulfur atom commonly serves as a bridge between other atoms in biochemistry, while an iodine atom is usually a terminal atom in a long molecule), but in that case, the same is true for other diatomic nonmetals---nitrogen and oxygen.--R8R (talk) 18:42, 28 January 2016 (UTC)

(another arbitrary break)[edit]

Thanks for those answers. Could you elaborate your WP:OR and/or WP:V concerns? Sandbh (talk) 10:35, 29 January 2016 (UTC)

What I am most afraid of is, I hadn't seen a non-Web table that puts "diatomic nonmetal" and "polyatomic nonmetal" as categories for a general classification before we adopted them in-Wiki. And a brief search in Google Books seems to show the same now, if one excludes books that are targeted at beginners at chemistry.--R8R (talk) 07:11, 30 January 2016 (UTC)

I haven't seen such a non-web table.

Some authors that come more or less close, in terms of concept or boundary lines, follow.

Fernelius and Robey (1935, p. 62) include a periodic table that divides the elements into four main classes, on the basis of crystal structure: I. The true metals (Groups 1 to 11); II. Metals with modified structures (Zn, Cd, Hg | B, Al, In, Tl | Pb); III. Elements with 8–N structures (Ga | C, Si, Ge, Sn | As, Sb, Bi | Se, Te, Po | I, …); and IV. the rest of the elements (N | O, S | H, F, Cl, Br | the noble gases). P is shown as belonging to both class III and class IV.

Wulfsberg (1987, p. 159) divides the non-noble nonmetals into "very electronegative nonmetals" (N | O | F, Cl, Br) and "electronegative nonmetals" (H | Si, Ge* | P, As,* Sb,* Bi | S, Se,* Te,* Po | I, At). [He refers elsewhere to the asterisked elements as metalloids, and to Ge, Bi and Po as "electronegative metals".]

DeKock & Gray (1989, p. 426) have a periodic table that categorises the elements into "metals only"; "intermediate structures" (B | C, Si, Ge, Sn | P, As, Sb, Bi | S, Se, Te); and "monatomic or diatomic molecules" (H | N | O |F, Cl, Br, I, At | noble gases). Just below this table there is an extract of groups 10—18 that has been disassembled into segments according to bulk coordination numbers. The segments from left to right are: metallic packing; three-dimensional networks (B | diamond, Si, Ge, gray tin); sheets or layers; tetrahedra; chains; rings; diatomic; atoms.

Birk (2005, p. 234) says in words, rather than a table: "The nonmetals typically exist as diatomic or polyatomic molecules, with the exception of the noble gases, which are monatomic. Bell and Garafalo (2005, p. 131) write, "This might be a good time to introduce the idea of diatomic elements (formulas containing a subscript greater than 2) and polyatomic elements (formulas containing a subscript greater than 2). Only seven elements are considered diatomic…and only a few are…polyatomic (such as S, P and C). Students might notice that the diatomic and polyatomic elements are located…above the stair-step line."

Silberberg (2006, p. 550) has a periodic table extract showing the structures of the representative elements. C is shown as "solid, covalent network"; P | S, Se | I as "solid, covalent molecule (diatomic or polyatomic)"; H | N | O | F, Cl as "gas, covalent molecule (diatomic or polyatomic)"; Br the same, but a liquid; and the noble gases as "single atoms"

It doesn't matter if you or I haven't been able to find a non-web periodic table using precisely this division. A periodic table is simply a graphic representation of what is written. Everything about our periodic table has been written before; it's just that no one (as far as we know) has shown it as a table. That's probably the same with many Wikipedia pages and other kinds of tables or pictures—they are mostly original, and simply draw together and represent what is said in the literature or seen in the world.

I haven't been able to find a non-web version of any of the iterations of our colour categorised table that predates when these first appeared (c. 2002) in Wikipedia. The closest I've been able to get to is in 1981 (Breck, Brown and McCowan, p. 149) and 1968 (Crawford, pp. 542–543). The Breck et al. table uses the following colour categories: metals | lanthanides and actinides | transition metals | semiconductors | life elements | halogens | noble gases. The earlier Crawford table uses black, light grey, vertical hatching, horizontal hatching, dark grey and mottled shading to distinguish: alkalis | metals | biogens | halogens | rare gases. Text labels appearing along the top and bottom-left of the table are "light metals"; "lanthanide series"; "actinide series"; "nonmetals"; and "rare gases".

After 2002 the closest non-web versions I've seen are "other nonmetal (or nonmetal) | halogen | noble-gas" type tables.

It would be a sorry day for Wikipedia if we were unable to show in tabular form what is written in the literature. But I don’t think that is what you intended to imply.

[I have not made up my mind yet. The discussion is helping.] Sandbh (talk) 05:10, 2 February 2016 (UTC)

  • Bell RL & Garofalo J (eds) 2005, Science units for grades 9-12, International Society for Technology in Education, Eugene, Oregon
  • Birk JP 2005, Chemistry, Houghton Mifflin, Boston
  • Breck WG, Brown RJC & McCowan 1981, Chemistry for science and engineering, McGraw-Hill Ryerson, Toronto
  • Crawford FH 1968, Introduction to the science of physics,, Harcourt, Brace & World, New York
  • DeKock RL & Gray HB 1989, Chemical structure & bonding", University Science Books, Mill Valley, CA
  • Fernelius WC & Robey RF 1935, ‘The nature of the metallic state’, Journal of Chemical Education, February, pp. 53–68
  • Silberberg M 2006, Chemistry: The molecular nature of matter and change, 4th ed., McGraw Hill, Boston
  • Wulfsberg G 1987, Principles of descriptive inorganic chemistry, Brooks/Cole Publishing Company, Monterey, CA
Hmm. I have carefully read your posting, but I have one important question regarding it. Does the following statement adequately summarize it, and if not, what's the mistake/what am I missing: "There has been a quite high number of classifications; regardless of them, it is not crucially important, because an exact categorization is not a fact that we should quote/present as widely accepted"?--R8R (talk) 12:27, 3 February 2016 (UTC)
I would say that the alkali metals, alkaline earth metals, Ln, An, transition metals, metalloids, halogens, and noble gases are widely accepted. And that there is no consistent treatment of the remaining metals and nonmetals. Sandbh (talk) 05:33, 5 February 2016 (UTC)
Thanks for the answer. I absolutely agree on each of these categories except I'll need a little time to rethink the halogens. (For example, "alkali metals" is an uncontested name for the category that includes the group 1 metals; I'm not yet sure about "halogens," because there may be other categories that include the group 17 nonmetals.)--R8R (talk) 07:27, 5 February 2016 (UTC)

S compared to I[edit]

Property More metallic Less metallic
Physical Melting point is marginally higher (115.21 v 113.7 °C); boiling point is substantially higher (444.6 v 184.36)
Ductility is better (as plastic sulfur)
Higher bulk coordination number (2 v 1+)
Band gap is bigger (~2.5 v ~1.2 eV) hence iodine has a metallic appearance, partially delocalised bonding, is a semiconductor in the direction of its planes, and has better electrical and thermal conductivity
Liquid iodine is an ionic conductor; S only becomes a liquid semiconductor at 900 °C i.e. some 455 degrees above its boiling point
Chemical Electron affinity lower (200 v 295 kj/mol)
EN lower (2.58 v 2.66)
Ionisation energy lower (10.36 v 10.45 V)
Most stable oxidation state is +6, versus –1 for iodine
Standard reduction potential is lower (+0.14 v +0.54 V)
The S2- sulfide ion, to be present in aqueous solution, requires highly alkaline conditions (> ~pH 13) whereas the I cation is common in aqueous solution across pH 1–14
— —

Analysis: In its physical properties iodine is marginally(?) more metallic than sulfur whereas sulfur is appreciably more metallic in its chemical properties. Sandbh (talk) 01:44, 7 February 2016 (UTC)

About this post: I understand this is towards answering "do S, I belong in a non-metal category?". As such, it is a subtly different topic (i.e., off-topic): it would not redefine our set of categories. It would only change the membership lists! Other topic here may have this same deviation. Now I don't mind that much, but it might become confusing & complex at some point. By then, I might warn again ;-). -DePiep (talk) 10:54, 7 February 2016 (UTC)
Oh, I posted this comparison to address R8R's earlier comment in this thread about distinguishing between iodine and S. I was also wondering how iodine, which I've referred to as the most metallic of the diatomic nonmetals, compared to sulfur (which I've referred to as the least metallic of the polyatomic nonmetals). Sandbh (talk) 11:15, 7 February 2016 (UTC)
Sure, these things happen. But still, it expands & distracts this --already big-- topic. One could consider splitting off these OT subsubthreads into a new section. To maintain overview. -DePiep (talk) 21:04, 8 February 2016 (UTC)
There exist many scales of electronegativity. By the Pauling scale, S<I, but by the Allen scale, S>I. All-carbon persulfuranes are known, and even S(CH3)6 has been hypothesized, so there is some evidence that S is less electronegative than C (the Pauling and Allen scales state the opposite). Do you know anything about the existence of all-carbon periodinanes? Is I(CH3)3 possible? Burzuchius (talk) 12:23, 9 February 2016 (UTC)

H compared to S[edit]

Property More metallic Less metallic
Physical — — Melting point (−259.16 v 115.21 °C) and boiling point (−252.88 v 444.6) are lower, hence gaseous
Nil ductility
Smaller bulk coordination number (1 v 2)
Greater band gap (15 v 2.2 eV) hence colourless, and less electrical and thermal conductivity
Chemical Electron affinity lower (73 v 200 kJ/mol)
Electronegativity lower (2.2 v 2.58)
Standard reduction potential lower (0.0 v +0.14 V)
Ionisation energy higher 13.6 v 10.36 V
Homoatomic hydrogen cations not capable of independent existence(?), other than in the gas phase

Analysis: S is appreciably more metallic than H physically; H is somewhat(?) more metallic than S chemically

Combine lanthanides and actinides into one category[edit]

I think Nergaal also suggested a while ago combining the lanthanides and actinides, but I would again not support this as there is not really a clear term for what they are together because of Lu and Lr spoiling everything by being d-block elements. "Inner transition metals" would the best if that word had ever actually been well-defined. Double sharp (talk) 03:59, 6 January 2016 (UTC)

"Lu and Lr spoiling everything by being d-block elements" - this is from an self-wrongfooting presumption. Whether an element is in a category is not prevented by/required for its block membership. iow, categories are not limited or set by block limits. Categories are not a subdivision of blocks, they are a subdivision of the metal-metalloid-nonmetal trend. If I'm correct, the same cross-block category issue is in play wrt group 12 and He. (The same rule applies to category-vs-groups: group membership does not forbid category membership; eg At/halogen, and most main group elements). -DePiep (talk) 08:25, 7 January 2016 (UTC)
What this is looking for is a name for the lanthanides and actinides combined into one category. The problem is that nobody seems to agree what an "inner transition metal" is (and by the IUPAC definition, if taken literally, uranium is a transition metal as it has d electrons that are lost when forming ions, even though I don't think anyone thinks this anymore, so even the TM definition is disputable). The term is not exactly well-defined, and there is not even a IUPAC definition to help us decide which to be primary. "f-block elements" would have been an elegant solution if not for Lu and Lr being in the d-block, which spoils it. That's what I meant. Double sharp (talk) 08:55, 7 January 2016 (UTC)
I see. But isn't this problem created by wanting to use "ITM" for the category? That would be a redefinition of the Ln+An into a problematic definition. As you describe it, ITM is block-based (and so is what I pointed out). If we would just combine Ln and An into one category(-color) and name it "Lanthanides and actinides", no problem exists.

Re collapsing the side discussion: no problem with me.

I think "inner transition metal" should not be dismissed simply because of its under-definiteness: What is "lanthanide"? (15 elements? 14, and if so, which ones?) "Transition metal"? (Sc/Y in or out? La/Ac? Lu/Lr? Zn/Cd? Hg?) A much more serious problem for me is that "lanthanide" and "actinide" are much better-known than "inner transition metals." Names are crucially important for our categorization. Most people who know the very basics of chemistry have at least heard the word "lanthanide" (or seen it in the PT hanging on the classroom wall). "Inner transition metal" is fine, I think, as a term, but much fewer people heard it before.--R8R (talk) 06:32, 3 February 2016 (UTC)

How come we have horizontal borders between those categories Ln/An/eka-An? are these periodic differences actaully category differences (metallic properties, ...)? Or is it just an old historic habit (because discovery-era differ)? If we group Ln+An together in one category, then eka-An is just an extension of that group. (So can have one color, be it regular or lighter for predicted elements). Time to ask in the other category section (on this page)?--DePiep (talk) 09:16, 3 February 2016 (UTC)
Depends on how you put it. The very term "actinide" was invented as an analogy to "lanthanide." Later elements of both categories have similar chemistry. If you ask me for non-Wiki categorization purposes, I'd, however, still keep them separated. One striking difference is that all lanthanides (except one, but for this reason it's not very important for the category, as little is known about it) are stable elements, while all actinides are radioactive (starting with thorium and uranium through short-lived transfermiums). The most characteristic actinides (the ones that we know the most about and that are most readily available) are quite different chemically from their lanthanides counterparts. For Wiki purposes, I'd absolutely object merging them.
Even if we combine the two, it doesn't mean superactinides will necessarily fall for that very color. The latter term also includes elements of the g-block, which are not found among the Ln or the An.--R8R (talk) 10:18, 3 February 2016 (UTC)

The extended table: get rid of the small period 9?[edit]

I have revisited the extended periodic table article. It does suggest, unlike our current extended PT, that the elements 169--172 will belong to the period 8, and not 9.

Pyykkö model
Displaced elements are in boldface
 8  119
9 165
  s-block g-block f-block d-block p-block

The two major sources, Fricke and Pyykko, both suggest the differentiating electron will be 8p, not 9p. As such, I propose we follow Pyykko's representation (see Pyykko), keeping the 8p3/2 elements in period 8, even if they come after the 9p1/2 elements.--R8R (talk) 22:16, 6 January 2016 (UTC)

Both Fricke and Pyykko have the same results, but they differ in how the present them in a PT. What we have now is Fricke's version. The Pyykko's representation puts electronic configurations above the periodic law-themed idea of how the elements can only follow in an ascending order. I like that: I believe electronic configurations are the reason why the periodic law works in first place. As such, since we have to choose between the two, the Pyykko's version makes more sense to me (and the period 10 does not have to start with 6g elements! and they would normally be in the period 9 anyway).--R8R (talk) 22:31, 6 January 2016 (UTC)
Let's turn to the eka-superactinoids. Fricke's version mentions elements 173-184 in a period 10. Pyykko's version, although not in the above template, would put them in the g-block of period 9, only ending with element 190. Any thoughts on whether to include the eka-superactinoids in the extended periodic table as a question separate from whether to use Fricke's version or Pyykko's version?? Georgia guy (talk) 22:50, 6 January 2016 (UTC)
That is, by the way, a great question. Strictly speaking, we don't have any data for any of these elements, except for a single element, 184, and even that data, quote Fricke (1975), a major source on the element, "must be regarded as a mere speculation," even if "plausible." The data comes from an older source from 1971, and Fricke questions the whole idea of the element 184, stating, aside from obvious problems with such a high atomic number, that "no calculations of the atomic behavior were available to them [authors of the 1971 paper]." It may be a great idea to stop right after the element 172.--R8R (talk) 00:45, 7 January 2016 (UTC)
Fricke does not place 173–184 in his periodic table, and there is still questioning about what on earth happens after 173, so I think stopping at 172 makes sense (and killing period 10).
Fricke vs Pyykkö is a tough one. People have used both. I personally prefer the Fricke one because it has 139 and 140 with the superactinides (like they're expected to be), and shows that despite not having the same n, 9p1/2 and 8p3/2 are expected to behave as though they formed one p-shell. I suppose you could say that it prioritises the predicted chemistry over exactly following orbitals. Both are right in a sense, but I get the feeling it is easier to talk about Fricke because then you don't need to contortedly explain what is going on with 139 and 140, and it keeps the elements in order of increasing atomic number. Double sharp (talk) 03:03, 7 January 2016 (UTC)
Personally, I like Pyykko better. But we are here to explain the idea to a wider audience, and we can't just rely on personal feelings. I think Pyykko would do the job better. One thing I would definitely not miss if I saw the Fricke table but didn't know about the whole theory behind it, why are there 20 elements in the g block? I even could draw the surprising (and false) conclusion of 20 g orbitals (surely I would look deeper if I had time, but what if not). I suspect many readers would not have the question, but only because they would not pay attention to that or be indifferent. Pyykko is easy to explain: An element is located in a cell corresponding to its differentiating electron. The heavier elements, unlike the known ones, may fill the orbital in an order different to one you'd expect from the lighter elements. Relativity. This common concept, in short, does summarize all irregularities, current or future. With Fricke, you have to go case-by-case, no p block in period 8, then an extremely short period 9, then period 10 starting with the g block (but again, I insist on considering stopping after element 172). These cases would be interrelated, but different still, and either way they are obviously particular cases, and the Pyykko version implies an extendable theory.--R8R (talk) 09:50, 7 January 2016 (UTC)
Now one more thing: While Pyykko does explicitly present his structure as a PT continuation, Fricke, in fact, does not. Next to each horizontal row in his scheme, Fricke puts names of shells being filled and a color scheme (a pattern scheme, actually) for them and not a number of another period. Nowhere in the text is it stated that these are the actual period 8 and period 9, rather than just used in an illustration-purposed graph (which illustrates filling of different shells in each element; the eighth row already had five shells filled, so a sixth one was left out for the next row. The other rows have no more than just four shells each, so such a break may be seen as desirable).--R8R (talk) 13:30, 9 January 2016 (UTC)
Actually he does say it in the abstract of the 1971 paper with Greiner and Waber: "The eighth chemical period ends with Z = 164 located below Mercury. The ninth period starts with an alkaline and alkaline earth metal and ends immediately similarly to the second and third period with a noble gas at Z = 172." Double sharp (talk) 14:55, 9 January 2016 (UTC)
Hmm, point taken, thanks for sharing. But that still does not negate my argument re representation.--R8R (talk) 15:22, 9 January 2016 (UTC)
True. But does it have to be the case that 8p3/2 must appear in row 8 just because it has n = 8? After all, 3d appears in row 4, because it fills between 4s and 4p, is a valence subshell with 4s in the 3d series when it fills, and because this way it keeps the elements in order of increasing atomic number (unlike what would happen if you slotted it in row 3 after 3p, with element 21 coming before element 19). Now consider 8p3/2. It fills after 9s and 9p1/2 and is a valence subshell with these two in the 8p3/2 series when it fills. If we fill it in row 8 right under 7p3/2, then reading the table row by row, we encounter element 169 before element 165, breaking the sequence of increasing atomic numbers, that Jensen treats as a requirement of the periodic law. This seems distasteful to me. Additionally distasteful to me is the fact that Pyykkö puts elements 139 and 140 as the representatives of 8p1/2, when this subshell has been a permanent fixture in the table since element 121, and has been full since element 127, and the fact that they are so far from elements 121–138, when there is not expected to be much of a chemical difference. At this point, with 5g, bits of 6f and 7d, as well as 8p1/2 all filling together, I would simply throw in the towel and just try to follow the atomic number order and chemistry, which results in Fricke's table. Double sharp (talk) 15:40, 9 January 2016 (UTC)
8p3/2 has to appear in the period 8, because for all blocks except the s block, n+l must be equal ns,p+1. this is commonly thought to be the basis for the periodic table. the s block is an exception, and the left-step table has been invented to avoid it, but it didn't hit the public. Jensen refers to a completely different issue; moreover, as a chemist and not a physicist, he may not even think about any continuation of the PT into elements 120+. "This seems distasteful to me." -- but we're not writing a music album review, we're writing an encyclopedia, this can't be a basis for any decision. As for Pyykko, he does have a reason: "For our electron book-keeping, we put the 8p* shell at E139 and E140. As seen in Table 2 for the previous element, E1385+, 6+, the available electrons will be placed in a full, 8s2 shell, the rest going to the 5g shell, and none yet to the 8p shell. This supports the present placement." The EC-twisting factor of the relativity must be taken in account. And as difficult as it is, EC is a more basic property than chemistry, and therefore it must be the basis for any table out there.--R8R (talk) 16:07, 9 January 2016 (UTC)
OK, what you say about n+l makes sense, but I'm not convinced about the ions argument. Consider Yb2+, which is [Xe]4f14. The outer electrons are in a filled 4f shell, and none are in the 6s shell. But that does not mean we should suddenly put 6s after 4f simply because it ionises first. Throughout the entire lanthanide series we have had 6s electrons, similarly to how throughout the entire superactinide series from element 121 we have had 8p electrons. Pyykkö himself notes the order as 5g ≤ 8p1/2, whereas he dares to put strict inequalities for the other subshells. I think all it proves is that 8p ionises first. Double sharp (talk) 16:16, 9 January 2016 (UTC)
Roughly speaking, this is an exceptional situation because it mixes ns and the next shell. This exception is another reason why the left-step table is such a big deal.--R8R (talk) 17:14, 9 January 2016 (UTC) make sense...
I suppose my only hangup left with this is that the expected chemistry is that E165 to E172 would behave like a short period 3. For example, you can see a smoothly rising trend in ionisation energies. In 3p, there is a break between p3 and p4 (the half-filled shell); by 5p the break in the trendline is still there, but now it monotonically increases; by 6p and 7p the break is at p2/p3, and we have a dip again (in 7p a huge dip that makes Fl and E118 similar in IE); and in 8p3/2+9p1/2 there is no dip at all. (This is because these two p-subshells are very similar in energy level; their electrons have almost identical eigenvalues, so the only effect – subshell splitting – is negligible.) Fricke connects the lines, so he clearly thinks it's a trend from E167 to E172.
So splitting them seems to obscure this relationship, and make E139 and E140 look like boron and carbon group elements when they should behave more like superactinides. Looking at Pyykkö's provided ionic electronic configurations, it looks like 8p1/2 is lost in the early superactinides until E138, and then at E139 and E140 it starts to stay in the +2 ions. But doesn't that just mean it's moving towards being a core shell that is unconcerned with chemistry? If so it still would make more sense in my opinion to account for it earlier when it is playing an active role in chemistry, rather than at its last chances of doing so.
I suppose the two tables have different goals. Fricke is trying to emphasise chemistry and the Z sequence while Pyykkö is trying to emphasise his calculated electron configurations. Even today you can find both used together pretty often (here's an example from Pershina). I still think it's easier to cover the chemistry in sections like E119–120; E121–155; E156–164; E165–172. Otherwise you have to jump around to explain why the trends continue through E139 and E140 and from E168 to E169. Double sharp (talk) 03:58, 10 January 2016 (UTC)
Yes, "the expected chemistry is that E165 to E172 would behave like a short period 3" must be the reason why Fricke chose his way. However, I, following his own words, do not think this and his whole chemistry beyond element 120, while plausible, can be taken for granted. Electronic configurations are a simpler, more basic property. As we are going into predictions with a not so great degree of certainty, I find it would make most sense to choose the one that emphasizes that simpler property.--R8R (talk) 12:56, 13 January 2016 (UTC)
Makes sense: I think I've turned into a Pyykkö supporter, then. Double sharp (talk) 08:26, 19 January 2016 (UTC)

sub-tread re coloring the period 8/9 p-block elements[edit]

Another question: are we sure that E167–170 are all going to be metals? If this is going to be like period 3, then maybe it would not be so. The group oxidation state being among the most important is not absolutely prohibitive – +5 and +6 are important for phosphorus and sulfur respectively too. So maybe since Fricke says that period 9 would be like period 3, we should simply copy over the classifications from period 3 (167 PTM; 168 metalloid; 169 and 170 polyatomic nonmetals; 171 diatomic nonmetal; 172 noble gas), along with huge caveats. Double sharp (talk) 03:08, 7 January 2016 (UTC)

Another great question in this subsection. Pyykko barely touches the issue of elements 165--172; Fricke has a little more, but even he does not go into detail regarding the similarity between his period 9 and period 3, other than mentioning the same range of oxidation states. He does not say anything other than that, and we should not try to guess what he actually meant, because a) we may guess wrong, and b) I am certain he said only as much as he could be certain about. If metallicity, for example, was similar for elements 167--172 and 13--18, he undoubtedly would add the word "metallicity" into the phrase about the similarity. Unless we have sources specifically saying these elements would be metals/nonmetals (I only have Pyykko and Fricke '75 on my hands, both being available in the Internet, and it's been a while since I last visited the topic; do we?), we should color them gray. That would most closely match the spirit of WP:V.--R8R (talk) 09:50, 7 January 2016 (UTC)

Here's my attempt at gathering every little scrap of information Fricke gives us:

  • Density: seem more suggestive of metals. However, note that Fl (a rather normal PTM) is expected to be at 14 g/cm3, some way beyond its lighter congener Pb at 11.3, so most of this is simply because of how low down these elements are on the PT and may not really have much of a bearing on their chemistry: they only increase as much as they would be expected to from density trends. This is pretty inconclusive, but suggests that this can be safely ignored as a consideration.
  • Stable oxidation states: no doubt there are some less stable states that Fricke doesn't mention, if period 9 is so close to period 3 in terms of chemical behaviour, like +4 or ±2 for element 170 or things like that. The statement of the group oxidation states being the most stable is consistent with period 3 and not period 6 (for example), but some of this is because of the inert pair effect in period 6. Now period 9 is expected to suddenly have no inert pair effect. So while this may not prevent the onset of metallization, reaching the group oxidation state would not be a problem. I could imagine 170 as a hypothetical polonium that had no trouble reaching +6. (Speaking of polonium, given the successes with longer-lived curium isotopes at higher oxidation states, I would humbly suggest trying to do chemistry on 208Po or 209Po instead of the standard 210Po. Bagnall suggested the same thing too, but to my knowledge nobody's done it because it's more difficult to work with these isotopes.)
  • First ionization energy: certainly gives the impression of the trend doing an about-turn back towards period-3 levels. But a little inconclusive; look at the huge ionisation energies of Rg and Cn, and that apparently doesn't stop them from metallising.

Now what about 171, which Fricke proudly proclaims will be a halogen similar to iodine? I think the answer lies in the fact that he specifically says iodine and not, say, chlorine. Iodine is close to the metalloid strip and has incipient metallic properties. Thus we could reasonably expect the formation H(171) and the soft base 171, but we do not know how stable these would be. We could argue for metalloid, since the EA is within the range of variation of the known value for At; but the Goldhammer-Hertzfeld criterion would push it over the line (already it is predicted that At may act as a metal anyway, so 171 may well do so too). So maybe 171 could be simply a metal that is stable in the −1 state. We do not know how stable the 171 anion is going to be (only that it will be part of its chemistry, according to Fricke), and crucially, Fricke abstains from calling it a halogen, only similar to the halogens in the −1 state. That would be very odd if it was expected to be almost a carbon copy of iodine, so I think he is hinting that the EA does not tell the full story, because 171 is so much further down the periodic table (two more rows!).

Hence I to my surprise have come full circle to Sandbh's original position on 171; that it would be a metal with a significant chemistry in the −1 oxidation state (certainly stable enough to be reduced there by H+, but this is possible for the perhaps-even-metallic At as well). Furthermore, given 117's much lower EA and the fact that −1 is expected to be the least common part of its chemistry, I find myself to my surprise wondering if we could plausibly colour it as a post-transition metal. Fricke expects it to be semimetallic, but the fact that he uses this term makes me wonder if he means the physics definition, in which case 117 could be a semimetal and a PTM. It's not unheard of: α-Sn and Bi are like that.

My initial assignment of diatomic nonmetal to 171 was made on the basis of Fricke's first paper, which predicts chlorine-like behaviour (171 being a hard base, for example). He dials back this prediction to iodine-like behaviour in his next paper, so I think metalloid is a more reasonable description.

I think it's safest to assign them the most metallic plausible description, as it's difficult to imagine what could possibly prevent the onset of metallisation. At this high density and this far down the periodic table, the extrapolated Goldhammer-Hertzfeld criterion seems quite unable to stop the onset of metallisation for any of these elements save 172. Even in 1971, Fricke et al. refuse to call 117 and 171 halogens, only members of the halogen group (i.e. VIIA), while they do not dare to do likewise for 118 and 172 as noble gases. I think that by this point, it means that any remnants of nonmetallishness have vanished.

Now what about 118 and 172, the last two noble gases? 118 would be funny: it would be neither noble nor a gas. In 1974 Fricke dares to say that it would behave more like a "normal element", continuing the trend towards increasing reactivity down from Xe. Rn already shows cationic behaviour. The difficulty is that it would still probably be monoatomic, and so there is no way to recolour it now (there would have been if we reunified the other nonmetals into "reactive nonmetals" or "typical nonmetals", the latter excluding hydrogen). Already its fluorides would probably be ionic and non-volatile (reminds me of At and Rn), and it would be tremendously polarizable. Which means that an argument for reunifying the nonmetals would be to allow a more accurate characterisation of 118. Maybe we should change its colour!

Interlude: "typical nonmetal" is not ok for a category name. what are the criteria for "typicalness"? "typical" is a generic term, and it is meant to be a generic term, without any set of elements assigned to it.--R8R (talk) 15:25, 9 January 2016 (UTC)

172 would be a better noble gas, but for the fact that it would almost certainly be a solid. Fricke also calls 172 in 1974 a strong Lewis acid, so that it can donate a lone pair of electrons (and thus its IE must not be that high – in fact, the predicted value is closer to Rn than Xe). Thus fluorides and oxides are certainly possible (they are already possible in Xe). Thus we have 172 dialing back the trend towards Xe and Rn, so that we could still colour it as a noble gas. (The issue is a little moot using the current colour scheme.) Except that the category name is annoying.

BTW, I read Pyykkö and found an explanation of why he puts 165 and 166 in groups IA and IIA (not the d-block groups IB and IIB); because despite being d10s1 and d10s2, Cn2+ is expected to be d8s2 (implying 6d < 7s), while 1662+ is expected to be d10s0 (implying 7d > 9s). But is this not true for Zn, Cd, and Hg as well? OTOH he does neatly explain why the 7d elements are in those groups; because while 8s is inactive, 8p1/2 is not, and stands in its place! Thus 9s, which fills up at 165 and 166, is a new s-shell and a new period can start. (If not for this, I'd seriously consider a table with four extra superactinides that crams 119–172 all into period 8, that eliminates the blank spaces in Fricke's table.) Double sharp (talk) 14:28, 9 January 2016 (UTC)

P.S. Another nice tidbit from Pyykkö: it seems the current consensus is that the PT is going to end at 172 or 173. This 2011 paper explicitly says that the last element with a 1s shell outside the negative energy continuum is 173, and that this "yields the end of the periodic table". They cite Greiner on the negative energy continuum and resulting effects, but they appear to regard that anything beyond this will behave pathologically weirdly. I expect that the situation would quickly get out of hand, such that you wouldn't have your shiny new period 10 element for long enough even for the nucleons to arrange themselves into nuclear shells. They anyway expect that something will happen and that it will be prohibitive to the existence of the atom. They also give a predicted electron configuration for 173: [Usb] 9p1
. Pyykkö also calls 165–172 the last main group elements. So I think we should cut 173–184 off our periodic table. It appears to be the consensus outside one Philip Ball article – who still admits that the exact details of what happens past 173 are unresolved (and they don't look good). Nobody's placed 173 on a PT, and judging from what I've read so far it's not sure if 173 is going to be a thing or not, but 172 will be. The limit (calculated as 173.17) is fairly close to 173. Double sharp (talk) 14:37, 9 January 2016 (UTC)

I will need some time to write a substantial reply, but for now, I think we can declare elements 173+ should go away. We have enough data to say that just on the basic of core Wiki rules. I'll remove them now (even though I was prepared for them with a color for the eka-transactinides.)
Besides, how do we attract more attention to the nonmetal discussion?--R8R (talk) 15:30, 9 January 2016 (UTC)
We had no difficulty with that in 2013 when it was the only thing discussed and our productivity dropped down the rabbit hole. ^_^ Seriously, I don't think it'll attract much attention until we finish with the actually active recolouring and extended PT discussions.
BTW, to me "eka-transactinides" is the series after the superactinides, from 156 to 172. It makes sense since the transactinides appear to be defined as elements 104 to 120 (some will include 121, if they are Sc/Y/La/Ac proponents and don't think Ac is an actinide, nor La a lanthanide), and the superactinides follow as 121 to 155. 173+ would be eka-superactinides if Fricke's half-hearted predictions were right, but this seems to be getting more and more unlikely. (In fairness he does say – though not in the words I'm about to use – that the further he gets past element 120, the more grains of salt you should take his predictions with.) Double sharp (talk) 16:20, 9 January 2016 (UTC)
Of course he does abstain from calling 171 a halogen. He himself said this was just a preliminary calculation. It would require a great degree of certainty to say 171 was a halogen, which he probably did not have due to limited data. So this does not tell us 171 is actually more metallic; if he with his data could draw the conclusion himself, he'd do it. Scientists are generally people passionate with acquiring new knowledge. If they could safely tell, they would. Remember the clear results on elements 119 and 120 (it is agreed these will be typical, in a wider sense, s block metals; note there is no such agreement on 117, let alone 171). 172 is okay for the label of a "noble gas," even though it's a solid: nowhere does that name require absolute accuracy, being a name, unlike a label of being just a "gas," wich is completely different in not being a part of classification, but a verifiable fact -- state of matter.
Throughout the time I was writing ununseptium, I was nowhere close to the conclusion it was a metal. Not because it isn't one, but because scientists must've got their data w/o classification on their minds. We call Cn a metal and Fl a metal mostly because scientists agreed to call them metals. And we, following WP:V, have to follow. I think it would be most correct (given that very pillar of Wiki, WP:V) to abstain from any declaration in situations when we ourselves can't come to an agreement, when our positions change over time. if we were a personal site, it would be OK, but we're writing an encyclopedia here, and all of our data should be as accurate and verifiable as possible. We should re-introduce gray to our extended PT until we got enough data to get rid of it for sure.--R8R (talk) 16:11, 15 January 2016 (UTC)
Not a bad idea at all. We would only need it for E167–171, because he explicitly states it for everything else. Double sharp (talk) 16:06, 16 January 2016 (UTC)
As for being a gas, I'm sure everything can be a gas if hot enough. Are there any elements that cannot be gases even at a very high temperature (write it in degrees Kelvin)?? Georgia guy (talk) 16:41, 15 January 2016 (UTC)
Of course not, but normally when we say "gas", we are talking about elements that are gases at standard conditions. Double sharp (talk) 16:04, 16 January 2016 (UTC)

Turn scandium, yttrium, and present lanthanides into new "rare earth metals" category[edit]

It has been proposed before, so I figure it deserves a mention. Double sharp (talk) 12:26, 11 January 2016 (UTC)

I posted about this at Periodic Table: Talk, here. Sandbh (talk) 09:59, 14 January 2016 (UTC)

The group 4–11 focus[edit]

Move group 12 over to the post-transition metals[edit]

Exactly what the title says (perhaps even including Cn till we get the experimental evidence otherwise). Both this and the group 3 issue are related to Sandbh's topic on the group 4–11 focus. Double sharp (talk) 16:08, 16 January 2016 (UTC)

"(perhaps even including Cn till we get the experimental evidence otherwise)" -- why not leave as is till we get the experimental evidence otherwise?--R8R (talk) 17:11, 16 January 2016 (UTC)
Currently we have physical properties, but not chemical properties. The physical properties are like the lighter group 12 elements; it's the chemical properties that are expected to be different, but we don't know for sure yet. Double sharp (talk) 05:29, 17 January 2016 (UTC)
I have re-read copernicium. It says copernicium is agreed to be a metal and a group 12 member, and nothing else.
The two alternatives---group 12 belongs to the TMs or the PTMs---are equivalent; I remember one of our articles (group 12 element or maybe transition metal) once quoted a 2003 secondary research saying both are evenly distributed among authors; half the authors marked them a TMs and another half as PTMs. Given that, the two versions are equivalent, and I won't advocate for either. But the current version easily allows grouping Cn as TM, and the other one does not allow grouping as any as of now. Physical properties tell us Cn is a group 12 element. If we accept that is enough for being a TM, question solved. But if we don't, then we only know it is a metal and a group 12 member; we don't yet know if the transition is over in period 7. And if we choose this scheme, then I demand Cn can't yet belong to either category and stay uncolored. (Which I wouldn't want, so if I was to choose, I'd leave it as is.)--R8R (talk) 11:17, 17 January 2016 (UTC)
I am tempted to then simply leave this particular issue on hold until we finally get experimental confirmation of Cn(IV). It strikes me that the group 4–11 definition is meant primarily for the stable elements. Then we can talk about atypical compounds like the one Hg(IV) compound known. But for the superheavies, even their fleeting existence is atypical, and it is not surprising that such definitions break down. And if we are leaving group 12 in the TMs, I suppose it makes sense to leave half of group 3 there as well. Double sharp (talk) 13:15, 17 January 2016 (UTC)
I agree; I think I'd do just that if I was to make the decision.--R8R (talk) 13:29, 17 January 2016 (UTC)

Article needed: Subcategories in Metallicity[edit]

On the topic of metallicity categories (ie, subcategories of the main metal-metalloid-nonmetal classification), we do not have an good article or section (descriptive, comparative). Since we spread these categories all over the world (by coloring our PT), such an article is desired.

  • Related articles:
Properties of metals, metalloids and nonmetals (good example, for this 3-set)
metallicity (in astronomy)
metal, metalloid FA, nonmetal GA
nonmetal#Categories, metal#Categories
  • Title issues
'Category' is our enwiki word (lack of international/IUPAC word for this). Even we use this ambiguously, see Metal (category: heavy metal etc).
-DePiep (talk) 10:03, 11 January 2016 (UTC)
See also the Metals, Metalloids and Nonmetals section in the Periodic table article, and Names_for_sets_of_chemical_elements, and Categorization. Sandbh (talk) 10:39, 11 January 2016 (UTC)
How about Metallicity of chemical elements? Lede sentence: "The elements of the periodic table can be categorized/characterized by their metallicity, that is, where they fall on the metal-nonmetal continuum."
Another interesting point that came out while working on Properties of metals, metalloids and nonmetals was the idea that transition metals are the "archetypal metals" and diatomic nonmetals are "essentially nonmetallic". Interesting perspectives I hadn't thought of before.
YBG (talk) 06:27, 12 January 2016 (UTC)
  • If I understand the recent thread by Sandbh well, this categorising (or grouping) is not just by metallishness and metallic properties, but also is supported by other chemical & physical properties. That makes the grouping stronger, but the decription as "category of metallicity" less applicable. The article is needed (even more!), but maybe with a different title. -DePiep (talk) 11:02, 14 January 2016 (UTC)

Content bonanza[edit]

The new chemistry of the elements Sandbh (talk) 09:48, 12 January 2016 (UTC)

  • Intriguing cover. Being a foreigner myself to the English The Royal Society, I might need to contact their Foreign Secretary, the honourable Martyn Poliakoff (of PTOV fame), to suggest that they could replace the 1946 Seaborg PT presentation with a correct 32-column form including some modern depiction of group 3. The PT the Secretary has on its wall for 50 years might be antique by now (in PT terms that is, not in TRS terms). -DePiep (talk) 10:58, 14 January 2016 (UTC)
If I'm not wrong I believe the cover has a 14LaAc (type 2) periodic table. Sandbh (talk) 05:02, 21 January 2016 (UTC)
Touch'e. -DePiep (talk) 01:17, 9 February 2016 (UTC)


I've added several sections to block (periodic table): §§ S-block, D-block, F-block, and G-block. The information is mostly gleaned from other WP articles with a bit of interpolation and extrapolation, so there are no external references and certainly one or more major inaccuracy. I would greatly appreciate having other sets of eyes take a look at it. Many thanks! YBG (talk) 08:54, 20 January 2016 (UTC)

When will we reach the end - of the periodic table?[edit]

An article from the Smithsonian Magazine. Note the 15LaAc (type 3) periodic table. Sandbh (talk) 05:00, 21 January 2016 (UTC)

Another example of using that PT structure without sourcing it. They've just copy-pasted an ill-adjusted wallchart, cramming in 32 elements in group 3. Tallying sources that make this mistake would win that argument, but not add any scientific base. Even more saillant is that you seem to see this as a proof of something.
Then there is this: One open question concerns lanthanum and actinium, which have less in common with the other members of their respective groups than lutetium and lawrencium. Er, "their respective groups" -- they say that La and Ac are in different groups? If, for understandings sake, we assume they mean to say "their respective category" (ie, Ln and An), this points to a different PT: they say La and Ac stand out, and so are likely to be in group 3. So they point to the group 3 = Sc/Y/La/Ac PT. -DePiep (talk) 09:08, 21 January 2016 (UTC)

Sc|Y|Lu|Lr table spotted in popular science media program[edit] on battery-powered homes, at 05:05. Sandbh (talk) 11:20, 2 February 2016 (UTC)

  • Watch this. At 0:30 the Sc/Y/Lu/Lr PT. And this is great: given the omission of names for element 110 and up, the design must have been quite old as Ds was named in 2003. (The Prof spoils it by, at 1:58, showing unmoved crippled oldfashioned Sc/Y/*/** PTs [1] more recently made by Theo Gray and ACS. Lot of work to do). -DePiep (talk) 15:14, 2 February 2016 (UTC)

Happy 182nd, Dmitri![edit]

Google honors Dmitri Mendeleev's 182nd birthday with a special doodle YBG (talk) 04:08, 8 February 2016 (UTC)

See also here YBG (talk) 04:10, 8 February 2016 (UTC)
It says: "the logic of his table (which reportedly came to him in a dream)". How did google find this? How to get rid of it? -DePiep (talk) 13:22, 8 February 2016 (UTC)
Is there any explanation on (a) the coloring used by google (did Mendeleev categorise elements like that)?, and (b) why 32S is singled out (a pivotal element)? -DePiep (talk) 13:36, 8 February 2016 (UTC)
The tale of how he first dreamed of it, rather than invented it the regular way is a legend, but people fall for such things. Many know as the truth. Why wouldn't they share the fun.
Take it as just a thing made for people's joy, not a scientific paper. I'd have my questions if there were a point in them -- but there's none, and I am just glad they paid tribute to Mendeleev and his table and are making another small step in popularizing science. If you can't, wait for it -- and tomorrow it's gone. No biggie. :) by the way, 32 is the element weight, not the atomic number -- there was no atomic number concept in Mendeleev's best years, and Google is correct on this.--R8R (talk) 14:31, 8 February 2016 (UTC)
'course. Questions same. -DePiep (talk) 15:10, 8 February 2016 (UTC)
Colors are aligned by columns because vertical relations between were crucial for creating the table in first place. The only thing that table could be criticized for is how K and Ca didn't get cokored as Na and Mg. Transition metals were in different groups than main group elemwnts, but the table showed only obe TM per column, so they didn't turn their PT into a Christmas tree. Remember, they were making a fun picture (why else would Mendeleev hold a cube labeled "32 S"), they are no chemists, they could pass the job exclusively to designers, they could make their mistakes.--R8R (talk) 07:24, 9 February 2016 (UTC)
  • Nice point-of-doodle indeed. But why would Google make such mistakes? These colors, I say it's huge. We should point Google to enwiki. -DePiep (talk) 01:16, 9 February 2016 (UTC)