In lieu of an abstract, here is a brief excerpt of the content:

20 Gosset: The Fable of the Bee Gosset came in to see me the other day. He is a very pleasant chap. Not at all the autocrat of the t table. yule, n.d., quoted in m. g. kendall 1952, 159 Given plenty of room bees rarely swarm. gosset to karl pearson, april 24, 1910 (folder 704/7, pearson papers, ucl) Karl Pearson couldn’t learn from Gosset. But Gosset couldn’t learn much from Pearson, either. “I am bound to say that I did not learn very much from his lectures,” he told Egon. “I never did from anyone’s and my mathematics were inadequate for the task,” he continued. “I had learnt what I knew about errors of observation from Airy” (i.e., from the astronomer G. B. Airy’s Theory of Errors of Observations [1861]). Gosset’s lecture attendance, though, was good, and “Student” was a diligent note taker—until one day he came upon the idea of z (as t was at first called) for small samples, at which point his notes on Pearson’s lectures go blank. Gosset thought his discovery “commercially important” on top of being scientifically correct, and he pursued it with all energy. He would later credit K. P. for “supplying the missing [mathematical] link in the probable-error-of-the-mean paper—a paper for which he [i.e., Karl Pearson] disclaimed any responsibility. I also learned from him how to eat seed cake,” Gosset said, “for at 5 o’clock he would always come round with a cup of tea and either a slice of seed cake or a petit-beurre biscuit, and expect us to carry on.”1 207 Fisher, unlike Pearson, did at first learn from Gosset. But his learning was of a selective and even monotypic kind and ultimately parasitic and manipulative in the style of a cuckoo. Gosset and Fisher became acquainted when the younger man and better mathematician wrote to Gosset at the Guinness Brewery in 1912. The letter and with it an enclosed technical note by Fisher concerned Gosset’s 1908 article on what Fisher would later call “tests of significance” for small samples—Gosset’s z-test.2 Fisher greatly admired Gosset’s article but pointed out a technical error: z should be divided by “degrees of freedom,” not sample size, a correction Gosset agreed with and which caused z to be renamed, in 1922, Student’s t.3 Gosset was impressed by the Cambridge boy’s ammendment. Impressive but difficult-looking maths, he thought. Then he lost the note—in, he feared, one of the northern lakes! Gosset, who was a sailor and fisherman and general outdoorsman, worked by managed chaos. His scientific notebooks resemble grandmother’s recipe books. One surviving notebook, for instance, contains a postcard from his brother Henry, a bill of receipt for sulphate of potash, a letter from Yule about some experiment they had previously discussed, and a formal memorandum from an Irish official —all scribbled over with Gosset’s calculations and aphorisms and stuffed between the notebook covers. Mail was to Gosset just another stack of scratch paper. His office at the brewery, Egon Pearson has said, was a pulsing mess.4 Eventually, it seems, Fisher mailed Gosset another copy of the note. Gosset was grateful for it, but it was still “mathematics of the deepest dye,” he told Karl Pearson. Gosset asked K. P. to publish Fisher’s correction as a “note” in Biometrika and he did so—which marked both the beginning and the end of K. P.’s harmonious relationship with Fisher (E. S. Pearson 1968, 445–8). Meantime, Gosset extended his usual courtesy to Fisher. By 1918 Fisher’s salutation “Dear Mr. Gosset” became in a letter the British-male-familiar “Dear Gosset,” and so remained in the 150 surviving letters to follow.5 The Bee Was the Teacher Gosset rarely depended on the test, even at the brewery (Ziliak 2008a). American hops preserved beer longer than Kent-raised hops, Gosset for example found—important to know when you’re marketing a natural, unpasteurized stout such as Guinness, shipped worldwide. The shelf life of stout stored in a cask, he found, using hops imported from midwest208 ⱐ The Cult of Statistical Significance ern American farms, was three to seven days longer than one using hops from Kent.6 But Gosset in this study and others often found z or t beside the point. “You want to be able to say ‘if farmers [or whomever...

Share