Talk:Red heat

«slightly»
That should read ‘very’: at 1000 °C the first table says ‘Orange’ but the second ‘Clear cherry red’ and at 1100 °C and 1200 °C the first table says ‘Yellow’ but the second ‘Deep / Clear orange’. So your temperature reading would be different by 100 K, or even 200 K.

Also, the figures in the second table come from Claude Pouillet 1836. 92.67.227.181 (talk) 03:07, 5 July 2022 (UTC)

I've consulted the source for the first table and it puts ‘orange red’ at the lower of the two values. Keeping that in mind the discrepancy shrinks considerably. It's also quite obvious that the strange °C figures are caused by a previous edition or source giving the figures in °F and that the article interprets the source in a messy way that wasn't intended.

Anyway, I've consulted seven different sources (Chapman, Pouillet, Halcomb, Howe, W&T, Ellern & Kemp) and compared them. For lower temperatures, the principal problem is terminology: how do I know that my cherry red is the author's cherry red? How dark is dark? Obviously, if an author's cherry is more orange than another author's cherry, it will also be lighter and he'll probably have a lighter dark red too. Pouillet gives no yellow, so with light orange he probably means a kind of amber. Howe skips orange... is his full yellow maybe orange or an orangey amber? Still, taking all of this in consideration, I think that below amber the differences in terminology don't reflect actual colour differences. Each of the colours is one that an author could plausibly use to describe the appropriate Planckian colour. (William Metcalf's remarks on the sloppiness of colour terms are particularly amusing.)

But in the amber to orange region, the temperature curve suddenly flattens. In this area, the differences between the authors are remarkably consistent: if the bend in the curve is in the amber region, their white will saturate near 1200 K, whereas if the bend is in the orange region their white will saturate near 1400 K. The only exception is Kemp which flattens much more smoothly, maybe because it's a pyrotechnical source.

The authors must have been basing their figures on materials with different emissivities. I'd like to integrate what I found in the article, but there are essentially two options: 1) Summarise all sources in one list of colours and temperatures, giving lower and upper bounds above amber. 2) Give all the sources explicitly. Because separate tables are hard to compare, they'd have to be put in a single table, but unfortunately that would mean throwing away the °F figures, because otherwise the table would become too unwieldy. — Preceding unsigned comment added by 92.67.227.181 (talk) 02:52, 7 July 2022 (UTC)


 * Also keep in mind that those are mere words.  So we're talking about e.g. different people's different meanings of the word "cherry red"  not differing attempts of trying to describe some fundamental reality.North8000 (talk) 13:11, 7 July 2022 (UTC)
 * That was what I meant with ‘an author's cherry is more orange than another author's cherry’. Have you read Metcalf's take on this? It had a nice understated sense of humour. 92.67.227.181 (talk) 14:35, 7 July 2022 (UTC)
 * I know that you already knew that. I was just pushing treating that more deeply. For example, sources would cover what people's definitions of cherry red are (or on maybe what definition a respected organization uses) but not on what cherry red is. North8000 (talk) 19:16, 7 July 2022 (UTC)
 * >sources would cover what people's definitions of cherry red are
 * Not that early on actually. The use of the word cherry as a red heat colour term got established long before things like Munsell, XYZ, Pantone or NCS were a thing. (Although it wouldn't matter, because iron and steel don't glow in the literal colour of ripe cherries so we know to take the name with a grain of salt.)
 * Or did you mean to say ‘sources would cover on which temperature they choose to stick the label cherry (and make different somewhat arbitrary decisions) but not what spectrum is emitted at that temperature’? Well, the emitted spectra are known. Iron's spectral emissivity is fairly flat (Hartsfield, Iverson & Baldwin) going from .55 at the red end via .57 to .64, which is why it looks grey. Same source contains a spectral reflectivity graph for gold and it's markedly different. Watanabe, Susa, Fukuyama & Nagata contains graphs for copper and silver if you're curious. This spectrum gets modulated by the Planck spectrum. (And the colour is further modified by the inability of the eye to adapt to extremely bright light but it seems normally that's no issue below amber, so that's irrelevant for our cherry picking.)
 * If I still haven't managed to touch upon the core of your point, the fault is probably all mine, but I'm a flawed human being so maybe help me out a little by being a bit more explicit if that's the case. 92.67.227.181 (talk) 23:45, 7 July 2022 (UTC)
 * It's probably so abstract and minor that I think that was my error to bring it up. But to pick a simpler case, it is not a scientific inquiry (or coverage) whether or not Pluto is a planet. Instead, it's a question/coverage about the definition of a mere word in the English language....."planet". North8000 (talk) 01:58, 8 July 2022 (UTC)

Possibly merge with Thermal radiation?
See my comment here: Talk:Thermal radiation 92.67.227.181 (talk) 14:43, 7 July 2022 (UTC)

The coloured table image on the right
The longer I look at it, the more I feel it cannot be right. Even in the complete absence of cone saturation, the red point should be at 1100 /MK or so which is about 900 K or 626.85 °C. Above this temperature the colour will inevitably acquire an orange tint. Now look at the table... Maybe it was created by someone who didn't know the difference between °C and K? Whatever the explanation might be, the image description doesn't say how the colours were arrived at. No sources, no formulae, nothing. 92.67.227.181 (talk) 16:21, 7 July 2022 (UTC)


 * I took it out. In Wiki terms it was challenged statements which are unsourced and unsourcable. North8000 (talk) 15:23, 8 July 2022 (UTC)

Thanks. I've tried to figure out what the colours should be, and even though I used a very crude calculation, I got much closer agreement with the sources I consulted: (I've made the background black to prevent glare.)  I'm not comfortable showing the full calculation yet because I got my constants mostly from Wikipedia and I've noticed in the past they can be subtly wrong. I don't think it'll matter for the end result but I just don't want to duplicate wrong figures. These were the steps I took: So that's how the sausage was made. This wasn't necessarily intended as a replacement for the old table that was removed, but primarily as a demonstration of how wrong the old table was. Note that the orange tint I spoke of is clearly present. — Preceding unsigned comment added by 92.67.227.181 (talk) 15:19, 11 July 2022 (UTC)
 * 1) I calculated the Planck spectrum.
 * 2) I modulated it by a linear approximation of the spectral emissivity in Hartsfield, Iverson & Baldwin. It made a difference but only a tiny one, so I think a more sophisticated calculation is unnecessary.
 * 3) I observed the spectrum with the CIE 1931 Standard Observer.
 * 4) I multiplied it by the matrix from CIECAM16 to get the cone-like responses.
 * 5) I clipped the maximum cone-like response to a fixed, somewhat arbitrary maximum. Yes, this is extremely crude, in reality the saturation will be more gradual. But I found a value for which I got a yellow and a creamy white at appropriate temperatures. Taking in consideration that at high temperatures different sources can be off by 100 K in either direction, maybe there is no point in trying to do better? After all, how are you going to determine how gradual the saturation should be if even the crude approximation is in close agreement with the sources?
 * 6) I transformed the result to XYZ and clipped it to the sRGB gamut by reducing the chroma. So the colours you see here are just a bit less saturated than the real ones. Although if you're working metal outside, the opposite might be true because the base grey of the metal will desaturate the glow colour. These colours are as viewed in darkness.
 * 7) I set the brightness of 700 °C to the sRGB maximum. For all the brighter colours I had to clip the brightness, because within the range considered, the luminance scales roughly as temperature to a power of 16.7.
 * 8) *So for every temperature doubling, the luminance goes up by more than a factor of fifty thousand. This is impressive, but in principle still within the human eye adaptation range, although how good its colour fidelity is going to be near the edges of its range is anyone's guess. Only about a third of luminance increase seems to be caused by the metal actually getting physically brighter, the rest being due to the spectrum shifting from infrared to the visible spectrum.
 * 9) *You may notice that for 426 °C the colour is only 1 sRGB red level; that wasn't intentional, it just worked out that way. Maybe it would be better fix the brightness of 700 °C to maximum instead? Then you'd maybe be able to see some more nuance in the darker colours.
 * Nice! My own preference would be degrees C & F. North8000 (talk) 20:02, 11 July 2022 (UTC)