# Flame Tests and Spectra

Laplace’s Mini World · Plate 75 · Matter, Light & Sound · Matter

> Put a little salt into a flame and the flame takes a colour. Spread that light with a prism and each metal shows its own few bright lines.

- Interactive model: https://world.lapla.si/en/flame/
- Embed: https://world.lapla.si/embed/en/flame/
- 中文: https://world.lapla.si/zh/flame/
- Updated: 2026-11-01

## 01 A Wire and a Bead of Salt

Dip a fine platinum wire into a metal salt and hold it in a nearly colourless gas flame, and the flame takes a colour: lithium crimson, sodium yellow, potassium lilac, calcium brick red, strontium crimson, barium apple green, copper green or blue.

In 1860 Kirchhoff and Bunsen did just this. They bent the end of a very fine platinum wire into a small eye, melted a bead of the dried chloride onto it, and put it into the edge of the flame from the side.

Chlorides are used because they evaporate easily. The seven dishes on the board hold the chlorides of these seven metals.

*Lithium crimson · sodium yellow · potassium lilac · calcium brick red · strontium crimson · barium apple green · copper green or blue*

## 02 Colour Belongs to the Metal

The colour comes from the metal, not from the salt’s other half and not from the fire. Kirchhoff and Bunsen put bromides, iodides, hydroxides, sulphates and carbonates of the same metals into seven flames: sulphur, carbon disulphide, alcohol, coal gas, carbon monoxide, hydrogen, and oxygen with hydrogen. Not one line moved; a different salt only changed how bright the lines were.

Their conclusion: the bright lines are a sure sign that the metal is there. The light comes from the metal’s atoms in the flame, not from ions in solution; a sodium ion absorbs and gives off light only far in the ultraviolet.

Calcium, strontium, barium and copper also form small molecules with oxygen, hydrogen and chlorine in the flame, such as CaOH, SrOH, SrCl, BaCl and CuCl. These give off bands of light, not fine lines, and most of these metals’ colour comes from the bands. It is still something made with the metal that shines.

*NaCl · Na₂SO₄ · Na₂CO₃: the lines in the same place*

## 03 A Fixed Amount of Energy

In the flame, sodium atoms are hit again and again by other molecules. One hit can lift the outermost electron from the lowest level, 3s, to the 3p level. On average it stays there only about 16 nanoseconds before it drops back to 3s and gives off the extra energy as one packet of light.

The 3p level is really two levels close together, 2.104 and 2.102 electronvolts above 3s, so sodium gives off light of only two kinds: 589.0 and 589.6 nanometres. By E = hc/λ, each packet of light at 589.0 nanometres carries 2.10 electronvolts.

Every element has its own levels, so it can give off light of its own few kinds only. Hydrogen’s levels and lines are on the page “Where Is the Electron?”.

*Sodium 3p → 3s · 589.0 and 589.6 nm · about 2.10 eV a packet*

## 04 A Flame That Hardly Shines

To see a salt’s colour, the flame should give off as little light of its own as it can. Most of a candle’s light comes from glowing specks of soot in the flame; without them a flame is just as hot but gives off very little light.

Bunsen and Roscoe described the lamp in 1857. The gas leaves a jet inside a metal block and draws in air through four holes 7 millimetres across; the mixture burns at the top of a tube 76 millimetres long and 8.5 wide, with a flame almost without colour and free of soot. Close the holes and the bright, smoky gas flame is back. The lamp had been in use in Bunsen’s laboratory in Heidelberg since 1855, made by the university’s instrument maker, Peter Desaga.

Kirchhoff and Bunsen wrote that the hotter a flame is and the less it shines, the clearer the lines, and that Bunsen’s lamp was very hot and gave very little light. The bead goes into the outer flame, a little above the inner cone.

*1857 · tube 76 mm long, 8.5 mm wide · four air holes 7 mm across*

## 05 Slit, Prism, Telescope

A spectroscope spreads a flame’s light out by wavelength. The light first passes a narrow slit at the focus of the lens in the first tube, the collimator, and comes out as a parallel beam. In the prism, the shorter the wavelength, the more it is turned. The second tube, a telescope, brings each wavelength to its own image of the slit: one wavelength, one bright line.

In 1860 Kirchhoff and Bunsen used a box blackened inside, with two small telescopes in two side walls set at about 58° to each other, and between them a hollow 60° prism filled with carbon disulphide. Their observing telescope magnified only about four times.

In 1861 the Steinheil workshop in Munich built them a new instrument, the kind on the board: a brass plate on a cast-⁠iron foot carries a 60° flint-⁠glass prism and the collimator, and two arms that turn carry a telescope magnifying eight times and a scale tube. The scale is a photograph of a millimetre rule, shrunk to about 1⁠/⁠15 on glass; the prism’s front face reflects it into the telescope, over the lines. Worked out for one flint glass, sodium’s yellow is turned about 48.2°, violet at 404 nanometres about 51.3°, and red at 766 nanometres about 47.2°.

*Slit → collimator → prism → telescope · sodium’s yellow turned about 48.2°*

## 06 Each Metal, Its Own Lines

Lithium gives one bright red line at 670.8 nanometres and a very faint orange one at 610.4. Sodium gives two yellow lines close together, at 589.0 and 589.6. Potassium gives a pair of deep red lines at 766.5 and 769.9, almost at the edge of what the eye can see, and a pair of violet lines at 404.4 and 404.7.

Calcium, strontium and barium mostly give bands. Calcium’s are orange and yellow-⁠green, at about 623 and 554 nanometres (CaOH); strontium’s are a spread of red (SrOH and SrCl, about 606⁠–⁠686) with one blue line at 460.7; barium’s are green bands (BaOH and BaCl, about 487⁠–⁠532). Copper’s blue comes from CuCl, at about 428⁠–⁠452.

Kirchhoff and Bunsen wrote that where a line falls is a property as unchanging as an atomic weight, and can be measured almost as precisely as in astronomy.

*Lithium 670.8 · sodium 589.0⁠/⁠589.6 · potassium 766.5⁠/⁠769.9 · strontium 460.7 · calcium 422.7 · barium 553.5 (nm)*

## 07 Yellow Over Everything

Sodium is almost everywhere. Kirchhoff and Bunsen found that a platinum wire burnt clean showed the sodium line again after a few hours in the air, and that tapping a dusty book made the sodium line flash in a flame several paces away. When they burnt 3 milligrams of a sodium salt in a corner of a room of about 60 cubic metres, the sodium line in a flame in the room took about 10 minutes to fade away completely, though the sodium reaching the flame was far too little to weigh.

Sodium’s yellow also falls where the eye is very sensitive. On the CIE’s daylight curve, the same power at 589 nanometres looks about 20,⁠000 times brighter than potassium’s 766 nanometres (computed here). So a little sodium in a potassium salt turns the lilac flame yellow to the eye.

Chemistry books look through two thicknesses of cobalt blue glass: the glass absorbs sodium’s yellow and the potassium flame shows red. A spectroscope does better, with each light in its own place. In 1860 Kirchhoff and Bunsen put six chlorides into a flame together, no more than 0.1 milligram of each; the lines of sodium, lithium, potassium and barium came one after another, and last those of calcium and strontium, as if out of a mist.

*At the same power, 589 nm looks about 20,⁠000 times brighter than 766 nm*

## 08 Two Unknown Blue Lines

In May 1860 Bunsen wrote to the Berlin Academy that the mother liquor of several salt springs held a fourth alkali metal. Several kilograms of it held barely a few milligrams, yet its spectrum was easy to spot: just two blue lines, one almost on strontium’s blue line and one a little further towards the violet, as bright as lithium’s red. No known element had two such lines there.

In 1861 they named it caesium, from the Latin caesius, which the ancients used for the blue of a clear sky. To get the few grams of salt they needed, they worked through about 44,⁠200 kilograms of salt-⁠spring water from Dürkheim, which yielded 7.27 grams of caesium chloride.

The same year, from about 150 kilograms of lepidolite from Saxony, they found a fifth alkali metal: two bright violet lines, and two red lines lying even beyond line A at the red end of the Sun’s spectrum. They named it rubidium, from rubidus, for that deep red. The flame alone cannot tell them apart: potassium, rubidium and caesium all turn a flame violet, and only the spectroscope separates them.

*Caesium 455.5 and 459.3 nm · rubidium 780.0 and 794.8 nm · about 44,⁠200 kg of spring water → 7.27 g of caesium chloride*

## 09 From a Flame to the Sun

In 1859 Kirchhoff let sunlight pass through a flame with salt in it before it entered the spectroscope. With weak sunlight, two bright lines appeared where line D should be; with strong sunlight, the two dark lines there grew deeper. His conclusion: the sodium in the flame absorbs light at the same wavelengths it gives off, and the dark lines D in the Sun’s spectrum show that there is sodium in the Sun’s atmosphere.

From then on the spectroscope was a tool for finding elements. In 1861 William Crookes saw a bright green line no one had seen before, in residues from a sulphuric acid works, and named the element thallium, from the Greek for a budding twig, for its fresh green. In 1863 F. Reich and Th. Richter saw an indigo line in impure zinc chloride from roasted ores and named the new metal indium. In 1868 astronomers saw a yellow line in the Sun’s prominences, close to D but matching no flame on Earth; Frankland and Lockyer proposed to call its element helium.

The sky has such lines too: the green and red of an aurora are light given off by oxygen atoms high up when they are struck (see “Curtains in the Sky”).

*Thallium 535.0 · indium 451.1 · helium 587.6 (nm)*

## 10 Flame Colours in the Sky

Fireworks use the same metals: strontium for red, barium for green, copper for blue, sodium for yellow, calcium for orange. Most of the light comes from small molecules formed in the fire: SrCl and SrOH for red, BaCl and BaOH for green, CuCl for blue. Traditionally, strontium and copper are paired with something containing chlorine, so that these chlorides form.

Blue is the hardest. CuCl, which gives the blue, is a fragile molecule in the gas: if the flame is not hot enough the light is too weak; if it is too hot, it breaks down into CuOH, which shines green, and CuO, and the colour washes out. One estimate puts the limit at about 1,⁠500 kelvin. So a firework’s blue is usually dimmer than its reds and greens.

The eye does not help: at the same power, blue light at 450 nanometres looks only about 1⁠/⁠26 as bright as green at 555 (computed here from the CIE’s data). The barium compounds formed in the burning are poisonous, and researchers are also working on green fireworks without barium.

*Red SrCl, SrOH · green BaCl, BaOH · blue CuCl · yellow sodium · orange calcium*

## 11 The Spectroscope, Taken Apart

The set-⁠up comes apart into seven pieces: the prism cover, the prism, the collimator, the telescope, the scale tube, the Bunsen burner and the wire holder.

Rest the pointer on a piece and it lifts out on its own, with its name; otherwise all seven are named beside them.

*Prism cover · prism · collimator · telescope · scale tube · Bunsen burner · wire holder*

## 12 Try Another Salt

This is the page’s model: where the lines fall follows NIST and the literature; their brightness and the flame’s colour are a drawing. Each press of “Dip again” puts a fresh bead of salt on the wire: the colour comes up, holds a few seconds, and fades as the salt is used up.

“A trace of sodium” adds a little sodium to the salt: the flame looks yellow, and sodium’s two yellow lines appear beside the salt’s own, which are all still there.

*The page’s model · positions from NIST and the literature · brightness a drawing*

**Try this**: In the lab, pick KCl and switch on “A trace of sodium”: what colour does the flame look now, and are potassium’s deep red lines still in the spectroscope? About how many electronvolts does one packet of sodium’s 589.0 nm light carry?

## What the model leaves out

The spectroscope on the board is a model after the 1861 paper’s description of Steinheil’s new instrument (its plate could not be found), in proportions taken from later instruments of the same kind, not a copy of a surviving one; Kirchhoff and Bunsen’s own of 1860 had two tubes and a hollow prism filled with carbon disulphide. In 1861 a black cloth over the prism and tubes kept out stray light; here it is a black cover, cut open at one corner, and the collimator and telescope have windows in their walls. The small lamp that lit the scale is left out. The burner follows the description of 1857. The light’s path from the flame to the telescope and the view in the telescope are drawings: the lines fall where NIST’s wavelengths and the published bands put them, but their brightness and width are only a guide, not measurements, and NIST’s relative intensities cannot be compared between elements anyway. The prism’s dispersion is worked out for Schott F2 flint glass, and the view’s scale is in nanometres, where the old instruments carried their own arbitrary scale. The flames are drawn in the colours chemistry books name, which a screen cannot show exactly. For copper only CuCl’s blue bands are drawn: no reliable wavelengths were found for CuOH’s green. Rubidium’s and caesium’s strongest lines are in the infrared; the view shows only what the eye can see.

## Easily misread

A flame test is not the metal burning, and the colour is not the salt’s colour or the fire’s: the flame’s heat excites the metal’s atoms (or small molecules containing the metal), and as they drop back they give off light at their own few wavelengths only. The light comes from atoms in the flame, not from ions in solution. A colour seen by eye is not a reliable test: a trace of sodium hides other colours, lithium, strontium and calcium all look red, potassium, rubidium and caesium all look violet, and copper can look blue or green; the position of the lines is what is reliable. Caesium’s sky blue is the colour of its lines; its flame looks violet. A firework’s colours are also light given off, not dyes.

## Common questions

### Is a flame test colour the metal’s, the salt’s or the fire’s?

The colour comes from the metal, not from the salt’s other half and not from the fire.

See: [02 Colour Belongs to the Metal](https://world.lapla.si/en/flame/#metal)

### What does a spectroscope do?

A spectroscope spreads a flame’s light out by wavelength.

See: [05 Slit, Prism, Telescope](https://world.lapla.si/en/flame/#scope)

### Why does a trace of sodium hide other flame colours?

Sodium’s yellow also falls where the eye is very sensitive.

See: [07 Yellow Over Everything](https://world.lapla.si/en/flame/#mask)

### Where does the name caesium come from?

In 1861 they named it caesium, from the Latin caesius, which the ancients used for the blue of a clear sky.

See: [08 Two Unknown Blue Lines](https://world.lapla.si/en/flame/#discovery)

### Why are blue fireworks the hardest?

CuCl, which gives the blue, is a fragile molecule in the gas: if the flame is not hot enough the light is too weak; if it is too hot, it breaks down into CuOH, which shines green, and CuO, and the colour washes out.

See: [10 Flame Colours in the Sky](https://world.lapla.si/en/flame/#fireworks)

## Sources

1. [G. Kirchhoff and R. Bunsen (1860): Chemische Analyse durch Spectralbeobachtungen, Annalen der Physik und Chemie 110 (the apparatus, the lamp, lines that belong to the metal, each element’s spectrum, the smallest amounts seen, a new alkali metal foretold)](https://archive.org/details/sim_annalen-der-physik_1860_110_6)
2. [The same in English: Chemical Analysis by Spectrum-⁠observations, Philosophical Magazine 20 (1860)](https://archive.org/details/londonedinburghd20lond)
3. [G. Kirchhoff and R. Bunsen (1861): the second paper of the same title, Annalen der Physik und Chemie 113 (caesium and rubidium: the material, the names, the lines)](https://archive.org/details/sim_annalen-der-physik_1861_113_7)
4. [R. Bunsen to the Berlin Academy, Monatsberichte 1860 (a fourth alkali metal)](https://archive.org/details/monatsberichtede1860knig)
5. [R. Bunsen to the Berlin Academy, Monatsberichte 1861 (the name caesium, a fifth alkali metal)](https://archive.org/details/monatsberichtede18611knig)
6. [G. Kirchhoff (1859⁠/⁠1860): Ueber die Fraunhofer’schen Linien, Annalen der Physik und Chemie 109 (sunlight through a salt flame)](https://archive.org/details/annalenderphysi157unkngoog)
7. [NIST: Atomic Spectra Database (line wavelengths, levels, transition probabilities)](https://physics.nist.gov/PhysRefData/ASD/lines_form.html)
8. [NIST: Explanation of the lines tables (relative intensities are not comparable between elements)](https://physics.nist.gov/PhysRefData/ASD/Html/lineshelp.html)
9. [NIST: CODATA 2022 recommended values (h, c, e)](https://physics.nist.gov/cgi-bin/cuu/Value?h)
10. [CIE: spectral luminous efficiency for photopic vision, V(λ)](https://cie.co.at/datatable/cie-spectral-luminous-efficiency-photopic-vision)
11. [H. G. C. Human and P. J. Th. Zeegers (1975): Molecular fluorescence of CaOH, SrOH and BaCl in flames, Spectrochimica Acta B 30](https://dspace.library.uu.nl/handle/1874/16421)
12. [R. W. B. Pearse and A. G. Gaydon (1941): The Identification of Molecular Spectra (band heads of BaCl, SrCl, CuCl)](https://archive.org/details/in.ernet.dli.2015.212925)
13. [J. Glück, T. M. Klapötke and T. Küblböck (2020): Fine-⁠Tuning: Advances in Chlorine-⁠Free Blue-⁠Light-⁠Generating Pyrotechnics, Eur. J. Inorg. Chem.](https://epub.ub.uni-muenchen.de/71295/1/ejic.201901097.pdf)
14. [J. Glück (2018): doctoral thesis, LMU Munich (pyrotechnic emitters, barium)](https://edoc.ub.uni-muenchen.de/22513/1/Glueck_Johann.pdf)
15. [M. A. Duncan (2025): Misconceptions and Insights about Flame Tests, J. Chem. Educ. 102 (the light comes from atoms, not from the ions in solution)](https://doi.org/10.1021/acs.jchemed.5c00142)
16. [OpenStax: Chemistry 2e, 18.1 Periodicity (flame colours of the alkali and alkaline earth metals)](https://openstax.org/books/chemistry-2e/pages/18-1-periodicity)
17. [Royal Society of Chemistry: Flame tests using metal salts](https://edu.rsc.org/resources/flame-tests-using-metal-salts/1875.article)
18. [Royal Society of Chemistry: Flame colours, a demonstration](https://edu.rsc.org/resources/flame-colours-a-demonstration/760.article)
19. [Royal Institution: Christmas Lectures 2012 teaching resource, Group 1 flame tests (rubidium, caesium)](https://edu.rsc.org/download?ac=13375)
20. [A. I. Vogel: A Text-⁠book of Macro and Semimicro Qualitative Inorganic Analysis, 4th ed. (the Bunsen flame, sodium masking potassium, cobalt glass)](https://archive.org/details/in.ernet.dli.2015.270728)
21. [J. E. M. Goldsmith (1990): What makes the colors in candle flames?, Optics & Photonics News](https://osa.magnet.fsu.edu/teachersparents/articles/pdfs/what%20makes%20the%20colors%20in%20candle%20flames.pdf)
22. [NPR (2013): Exploding The Mystery Of Blue Fireworks (interview with John Conkling)](https://www.npr.org/transcripts/198781855)
23. [P. E. Smith (2019): Red, white but rarely blue, The Conversation](https://theconversation.com/red-white-but-rarely-blue-the-science-of-fireworks-colors-explained-119284)
24. [Schott F2 flint glass, its dispersion formula (Schott data as given by refractiveindex.info)](https://refractiveindex.info/?shelf=glass&book=SCHOTT-F&page=F2)
25. [R. Bunsen and H. Roscoe (1857): Photochemische Untersuchungen, Annalen der Physik und Chemie 100 (the gas lamp, its principle and sizes)](https://archive.org/details/sim_annalen-der-physik_1857_100_1)
26. [The same in English: Photo-⁠chemical Researches, Part I, Philosophical Transactions 147 (1857), Plate XVIII](https://archive.org/details/per_abstracts-of-the-papers-printed-in-the-philoso_1857_147)
27. [W. B. Jensen (2005): The Origin of the Bunsen Burner, J. Chem. Educ. 82 (1855, Peter Desaga)](https://doi.org/10.1021/ed082p518)
28. [Science History Institute: spectrometer in the Bunsen–Kirchhoff style, 2008.047.002 (its form and size)](https://sciencehistory.pastperfectonline.com/webobject/F015DC38-84B9-4F86-9942-706297351262)
29. [Science Museum Group: burners and platinum-⁠wire stands for use with a spectroscope, 1880-225 Pt2](https://collection.sciencemuseumgroup.org.uk/objects/co8052775/bunsen-burners-and-accessories-1880)
30. [W. Crookes (1861): On the Existence of a New Element, Chemical News 3 (thallium’s green line and name)](https://archive.org/details/chemicalnewsjour03londuoft)
31. [F. Reich and Th. Richter (1863): Vorläufige Notiz über ein neues Metall, Journal für praktische Chemie 89 (indium’s indigo line)](https://archive.org/details/journalfurprakt00erdmgoog)
32. [J. N. Lockyer (1868): Notice of an Observation of the Spectrum of a Solar Prominence, Proc. R. Soc. 17](https://archive.org/details/proceedings1718681869roya)
33. [British Association report of 1871, the President’s Address, footnote (Frankland and Lockyer propose the name helium)](https://archive.org/details/reportofbritisha72brit)

## Data

Line wavelengths, level energies and transition probabilities from the NIST Atomic Spectra Database (ver. 5.12); the constants h, c and e are the exact CODATA 2022 values; the eye’s sensitivity V(λ) is the CIE 1931 photopic function (CIE 018:2019). The molecular bands in flames after H. G. C. Human and P. J. Th. Zeegers (Spectrochimica Acta B, 1975), R. W. B. Pearse and A. G. Gaydon (The Identification of Molecular Spectra, 1941) and J. Glück, T. M. Klapötke and T. Küblböck (Eur. J. Inorg. Chem., 2020). Flame colours after OpenStax Chemistry 2e, the Royal Society of Chemistry’s education pages, the Royal Institution and A. I. Vogel’s Qualitative Inorganic Analysis (4th ed.). The history after G. Kirchhoff and R. Bunsen, Chemische Analyse durch Spectralbeobachtungen (Annalen der Physik und Chemie 110, 1860, and 113, 1861), R. Bunsen’s letters to the Berlin Academy (Monatsberichte, 1860 and 1861) and G. Kirchhoff, Ueber die Fraunhofer’schen Linien (1859⁠/⁠1860); fireworks after J. Conkling (NPR, 2013), P. E. Smith (The Conversation, 2019) and J. Glück (dissertation, LMU Munich, 2018). The prism’s dispersion is computed for Schott F2 flint glass from its published Sellmeier formula; the photon energies and the comparison of brightness for the eye are computed here. The spectroscope on the board is this page’s model after the description of Steinheil’s instrument in Kirchhoff and Bunsen’s paper of 1861, in proportions from later instruments of the kind (Science History Institute, Science Museum Group), not a copy of a surviving instrument; the burner after Bunsen and Roscoe (1857). Material presets and the model kit as on the site’s other model plates (public/looks/CREDITS.md).

## Licence CC BY-NC 4.0

The text, pictures and models on this page are licensed CC BY-NC 4.0: free to use and adapt for teaching and other non-commercial purposes, with credit to world.lapla.si. Third-party material listed above keeps its own licence.

https://creativecommons.org/licenses/by-nc/4.0/deed.en

## Cite this plate

Laplace’s Mini World, Plate 75, “Flame Tests and Spectra”, https://world.lapla.si/en/flame/, CC BY-NC 4.0, updated 2026-11-01.