Thanks JL for saving him from "a disgraceful blunder". Following their conversation he has divided the New Zealand flora as JL suggested and finds genera with four or more species are more variable than those with three or less. It will take several weeks to go back over all his material.
My dear Lubbock
You have done me the greatest possible service in helping me to clarify my Brains. If I am as muzzy on all subjects as I am on proportions & chance,—what a Book I shall produce!—
I have divided N. Zealand Flora as you suggested. There are
339 species in genera of 4 & upwards & 323 in genera
of 3 & less. The 339 species have 51 species presenting one or
more varieties— The 323 species have only
37: proportionally (as 339:323 \:\: 51.:48.5) they ought to have had 48
I am quite convinced yours is the right way; I had thought of it, but
I am quite shocked to find how easily I am muddled, for I had before thought over the subject much, & concluded my way was fair. It is dreadfully erroneous. What a disgraceful blunder you have saved me from. I heartily thank you—
Ever yours | C. Darwin
It is enough to make me tear up all my M.S. & give up in despair.—
It will take me several weeks to go over all my materials. But oh if you knew how thankful I am to you.—
- f1 2123.f1Dated by the relationship to the letter to J. D. Hooker, 1 August .
- f2 2123.f2CD's calculations on J. D. Hooker 1853–5 are in DAR 16.2: 239a–46.
- f3 2123.f3Since 1855, CD had been engaged in recording the incidence of varieties in various botanical and zoological catalogues. He wished to show that varieties were most frequent in genera that also contained a large number of species; further calculations aimed to demonstrate that these large genera were also geographically widespread and their component species individually abundant. His calculations consistently gave him the correlations he hoped for, but he failed to notice that his results were the entirely artificial consequence of a faulty method that would always have indicated an apparent positive correlation between genus size and any chosen characteristic. CD's manuscripts (DAR 15.2, 16.1, and 16.2) show that in July 1857 he changed his method of calculation. Under Lubbock's guidance, he began to compare the proportion of variable species in large genera with the proportion in small genera using a valid method like the one described in the letter. See J. Browne 1980.