Tokyo 1908 · Washington 1912 · Yamaguchi 1967 · the receptor, 2002

Had We Nothing Sweeter Than Carrots

In 1908 the chemist Kikunae Ikeda pulled glutamic acid out of dried kelp and said its salts carry a fifth taste. Mix glutamate with inosinate, the substance in dried bonito, and the taste does not add, it multiplies. Here is Shizuko Yamaguchi's 1967 law for that multiplying, made operable and checked against two later measurements: it holds at small doses, bends at large ones, and is printed wrong in two of the places you would look it up. Then a century of English books deciding whether there are four tastes or five.

A taste named for convenience

Ikeda's paper appeared in the Journal of the Chemical Society of Tokyo in 1909, in Japanese; his patent on a glutamate seasoning had been applied for in April 1908 and granted that July. An English translation, shortened to three quarters of its length, was published in 2002. It opens with the physiologists of his day, who recognised four tastes and treated the rest as mixtures, and then disagrees with them:

“However, I believe that there is at least one other additional taste which is quite distinct from the four tastes. It is the peculiar taste which we feel as `UMAI [meaning brothy, meaty, or savory]', arising from fish, meat and so forth. The taste is most characteristic of broth prepared from dried bonito and seaweed [Laminaria japonica]. (…) I propose to call this taste `UMAMI' for convenience.”

K. Ikeda, J. Chem. Soc. Tokyo 30 (1909), 820 to 836; tr. Y. Ogiwara and Y. Ninomiya, Chemical Senses 27 (2002), 847 to 849. Brackets are the translators'.

He found the substance in the kelp by classical chemistry, crystallising it out of the broth, and identified it as glutamic acid; its sodium, calcium, barium and other salts all tasted of the thing he meant. He put the threshold at about one part in three thousand (“1 g of sodium glutamate or more is dissolved in 3 l of water”), and he offered a comparison that no one has improved on: “If these substances can be likened to color, `UMAMI' would be yellow and sweetness red.”

In September 1912 he brought the finding to the Eighth International Congress of Applied Chemistry in Washington and New York. The abstract printed in the congress volume does not say umami at all. It says: “For this taste quality the name ‘glutamic taste’ is proposed.” His speaking notes for that talk survive, in his handwriting and in a typed transcription the Umami Information Center publishes, and they contain the argument in its best form:

“An attentive taster will find out something common in the complicated taste of asparagus, tomato, cheese and meat, which is quite peculiar and can not be classed under any of the above mentioned qualities. It is usually so faint and overshadowed by other stronger tastes, that it is often difficult to recognize it unless the attention is specially directed towards it. Had we nothing sweeter than carrots or milk, our idea of the quality “sweet” would be just as indistinct as it is the ease with this peculiar quality.”

Ikeda, speaking notes for “On the Taste of the Salt of Glutamic Acid” (1912), p. 1, typed transcription, Umami Information Center. The transcription's spelling is kept.

Pure sodium glutamate was to be the honey and sugar of the new taste. The same notes say how hard it was to get: “about 30 grams of the acid was obtained from about 12 kilograms of the dried sea-weed”, though the best kelp holds about 1% of it, and that “two or three grams dissolved in one liter of water impart to it a very agreeable taste”. “Within the last three years”, the congress abstract says, its manufacture had arisen in Japan; the seasoning was registered in 1909 under the trademark AJI-NO-MOTO.

Two powders that do not add

In 1913 Shintaro Kodama found that the savoury substance of dried bonito, the other half of Ikeda's broth, is a different molecule: inosinate (IMP), a nucleotide. Decades later Akira Kuninaka, working on nucleotides from yeast, found guanylate (GMP) and the effect this page is about. The way Kenzo Kurihara tells it, Kuninaka tasted inosinate and found it weak, tasted glutamate and found it much stronger, then tasted the inosinate again without rinsing his mouth, and it was very strong: the inosinate had met the glutamate left on his tongue.

Shizuko Yamaguchi measured that effect with a tasting panel and published a formula for it in the Journal of Food Science in 1967. A mixture of u grams of MSG and v grams of IMP per 100 ml tastes as strong as y grams of MSG on its own, where

y = u + 1218 · u · v

The second term is a product. Nothing in it is a sum of the two tastes. Move the sliders.

Two things fall out at once. Glutamate alone scales as itself. Inosinate alone, by the law, tastes of nothing: set the MSG slider to zero and no amount of IMP moves the gold bar. That is not quite right in a glass of water, where inosinate has a faint savoury taste of its own, and Yamaguchi and Kumiko Ninomiya give the likely reason: saliva already carries about 1.5 parts per million of glutamate (in MSG equivalents), so the inosinate may only be enhancing the glutamate that is always in the mouth.

The law, as printed

The table under the bench gives the same mixture three times, because the law exists in three printed forms, and only one of them is usable.

• The working form above, y = u + 1218uv, in grams per 100 ml.
• Yamaguchi and Ninomiya's review “Umami and food palatability” (Journal of Nutrition, 2000), which is open access, prints it as y = u + γv with γ = 1218. We checked the page image; the u in the product is missing. Taken at its word, that form makes 0.01 g of inosinate on its own taste like 12 g of MSG per 100 ml, and it contradicts the review's own next sentence, which says inosinate by itself is weak. The same review's reference list dates the 1967 paper to 1970.
• The publisher's online record of the 1967 summary gives the product form correctly, with every quantity in g/dl, and then reads: “The γ value for IMP was calculated from extensive sensory testing as 1.218 × 10⁸.” In those units that constant makes a quarter of a gram of each in a litre taste like 761 kilograms of MSG per litre. Every later statement of the constant we found uses 1218, which is 1.218 × 10³. We have not seen the printed page of the 1967 summary, so we cannot say whether the 8 is in the print or crept in when the summary was digitised.

None of this changes the science. It changes what a careful reader gets if they look the law up and compute with the first version they find.

How many times stronger?

Accounts of the synergy give a multiplier, and they do not agree. Kurihara's 2015 review says a human's response to the mixture is “about 8 times larger” than to glutamate alone, and a rat's about 1.7 times. Zhang and colleagues (2008), reading a 2002 taste test, say 0.2 mM of inosinate made tasters 15 times more sensitive to glutamate. In cells carrying the human receptor, the same 0.2 mM shifted the glutamate needed for a half response 30-fold (Li and colleagues, 2002).

If the 1967 law holds, there is no single answer to give, because the product term grows with the amount. Hold the weight of the spoonful fixed and change only how it is split between MSG and IMP:

The best split is always close to half and half (exactly u = (1 + 1218c) / 2436 of MSG in a total c), and the gain at that split grows almost in proportion to the total, because the product of two halves grows as the square of the whole. At 0.01 g per 100 ml the best mixture is 3.6 times as strong as the same weight of MSG; at 0.05 g it is 15.7 times; at 0.1 g it is 31.0 times. A quoted multiplier with no concentration attached is not yet a number.

Where the law holds, and where it cannot

The law has a consequence that can be checked against measurements made by other people with other panels. Whatever the MSG level, IMP at v multiplies it by 1 + 1218v, so it should lower the threshold for tasting glutamate by the same factor. Two published thresholds let us try that.

At a small dose of inosinate, a formula fitted in 1967 predicts a 2002 measurement to within the roughness of a four-person test. At twenty-four times that dose it predicts a fall five times larger than the one measured. One row each is not a test of the law, and the two panels differ; but it is what a product law would do if the receptor behind it saturates, and Li and colleagues report that the effect of inosinate on the receptor is saturable.

Where would a product come from? Zhang and colleagues located both molecules on one part of the receptor, the “Venus flytrap” domain of the protein T1R1, and proposed that “L-glutamate binds close to the hinge region, and 5′ ribonucleotides bind to an adjacent site close to the opening of the flytrap to further stabilize the closed conformation.” Glutamate closes the trap; inosinate, which by itself does not switch the human receptor on, holds it shut. The standard way to write two molecules that help each other bind is an allosteric model, and the instrument below is that model: ours, the textbook algebra, not a fit to anyone's data. Its one knob, α, is how much tighter each binds when the other is present.

At low glutamate the signal in this model is the glutamate level times (1 + αi) / (1 + i), where i is the nucleotide in units of its own binding constant; while i is small that is 1 + (α − 1)i, which is Yamaguchi's form exactly, a glutamate term plus a glutamate-times-inosinate term. As the nucleotide rises the factor stops growing and levels off at α, and as glutamate rises the receptor fills and the multiplying stops. The model does not know the value of 1218; it only shows that a law shaped like Yamaguchi's is what cooperating binding looks like at the small end, and that the same binding forbids it from holding all the way up.

A century to be believed

“The scientific community received this discovery with moderate applause only,” Bernd Lindemann and his co-authors wrote in 2002. “Many, especially in English-speaking countries, remained unconvinced.” They name reasons: the full paper was in Japanese, umami is mild even at high concentrations, and glutamate never comes without a salty or sour partner. Yamaguchi and Ninomiya put the wait for international acceptance at more than 75 years; Kumiko Ninomiya, in 2015, dated the end of the argument to the human receptor papers of 2002.

English books keep a record of what their writers believed. The Google Books Ngram Viewer counts phrases in its scanned books by year; below are the phrases “four basic tastes” and “four primary tastes” against the same with “five”, in any capitalisation, and the word itself.

From 2014 on, “five” outnumbers “four” in every year. Before that it had led in only three years: 2011, and 1901 and 1926, down at the chart's floor. The word “umami” is near zero until the late 1970s, first passes “monosodium glutamate” above the chart's floor in 2009, and has outnumbered it in every year since 2012: the name of the taste now appears in books more often than the name of the chemical that carries it.

The chart has a floor, and the second view prints it. In the shaded years before 1908, when the taste had no name and MSG did not exist, “monosodium glutamate” and “Ajinomoto” (the name, registered as a trademark in 1909, under which Ikeda's seasoning was sold) should read exactly zero; they do not, most visibly in 1900. Those are books given the wrong date, which is common in a corpus this large. Treat anything near that floor as noise; the crossings above sit far above it.

What this page does not claim

It says nothing about whether MSG is harmful; The Syndrome That Couldn't Pass a Blind Test goes through the trials. Nothing here is a map of the tongue either (The Zones That Were Never There). And the 1967 panel's own concentrations, method and fit limits are in a paper we could not read in full: the curve above is the formula as its summary states it, drawn over a range we chose, not Yamaguchi's data.

Sources

1. K. Ikeda, “New seasonings”, tr. Y. Ogiwara and Y. Ninomiya, Chemical Senses 27 (2002) 847 to 849, doi:10.1093/chemse/27.9.847; from J. Chem. Soc. Tokyo 30 (1909) 820 to 836.
2. K. Ikeda, “On the Taste of the Salt of Glutamic Acid” (abstract), Eighth International Congress of Applied Chemistry, Original Communications vol. XVIII (1912), p. 147, archive.org. Speaking notes, handwritten and typed: Umami Information Center, umamiinfo.com/ikedakikunae, which also gives the 1908 patent dates.
3. S. Yamaguchi, “The synergistic taste effect of monosodium glutamate and disodium 5′-inosinate”, Journal of Food Science 32 (1967) 473 to 478, doi:10.1111/j.1365-2621.1967.tb09715.x (summary only; the full text is paywalled and was not read).
4. S. Yamaguchi and K. Ninomiya, “Umami and food palatability”, Journal of Nutrition 130 (2000) 921S to 926S, doi:10.1093/jn/130.4.921S. The printed formula, the saliva figure, the 75 years, and Table 3 (after Yamaguchi and Kimizuka 1979).
5. X. Li et al., “Human receptors for sweet and umami taste”, PNAS 99 (2002) 4692 to 4696, doi:10.1073/pnas.072090199.
6. F. Zhang et al., “Molecular mechanism for the umami taste synergism”, PNAS 105 (2008) 20930 to 20934, doi:10.1073/pnas.0810174106.
7. K. Kurihara, “Umami the fifth basic taste: history of studies on receptor mechanisms and role as a food flavor”, BioMed Research International (2015) 189402, doi:10.1155/2015/189402. Kuninaka's rinse, and the 8 and 1.7.
8. C. Sano, “History of glutamate production”, American Journal of Clinical Nutrition 90 (2009) 728S to 732S, doi:10.3945/ajcn.2009.27462F. The 1909 trademark.
9. B. Lindemann, Y. Ogiwara and Y. Ninomiya, “The discovery of umami”, Chemical Senses 27 (2002) 843 to 844, doi:10.1093/chemse/27.9.843.
10. K. Ninomiya, “Science of umami taste: adaptation to gastronomic culture”, Flavour 4 (2015) 13, doi:10.1186/2044-7248-4-13.
11. Google Books Ngram Viewer, corpus “en”, 1900 to 2022, no smoothing, captured 28 September 2026: books.google.com/ngrams.