382 MEMOIR OF AUGUSTUS DE MORGAN. 



1868. l L'ecole rationaliste commet la meme erreur dans la question 



de 1'origine du langage. L'homme parle parce qu'il a la 

 faculte, done il a pu in venter la parole.' 



The Schoolmen never generalised a quality until they had at 

 least two instances. As long as there was only A which had a 

 certain virtus, they said nothing about it ; it was occult, i.e. un- 

 known. But when B was found to have the same they had such 

 knowledge as comes of classification, being almost all they had. 



The moderns invented a name upon one instance, and made 

 it a cause. They said that magnetism was the explanation of the 

 magnet. The Schoolmen would have waited until the amber 

 showed its quality, and then the distinction of magnetism and 

 electricity would have been specific knowledge, the genus being 

 virtus attractiva. It is something to know two phenomena with 

 a generic agreement and a specific difference. 



If the medical candidate had known the mind of those who 

 classed, he would have said, I do not know ivliy except in that I 

 can refer the phenomenon to a class. We note agreements and 

 differences and arrange them. Arnauld, &c., might have a 

 similar answer made for them, but not for those who inferred 

 power of invention of languages from possession. 



To J. 8. Mill, Esq., M.P. 



September 20, 1868. 



MY DEAE SIE, Seeing you at Avignon again reminds me of 

 a question I intended to ask long ago. If a mathematician were 

 asked what Avignon reminds him of, I do not know what he 

 would answer, except the Avignon edition of Gardiner's Logarithms. 

 Gardiner published a very celebrated folio of logarithms in 1742, 

 with a very solid subscription list of 120 persons, of whom two- 

 thirds are now known in the history of science that is, to a 

 close inquirer. I greatly doubt that an old list could be found 

 except this of which the same proportion could be recovered. 

 Gardiner corrected all the errors with his own hand in all the 

 copies. My great-grandfather, James Dodson, who also in 1742 

 published his Anti-logarithmic Canon, with 1,100,000 computed 

 figures, did the same thing. I suspect there was some concert 

 as to this excellent plan between the two. 



To proceed, in 1770 appeared at Avignon the reprint of 

 Gardiner, in folio, ' Chez T. Aubert, Imprimeur, Libraire Hue de 

 1'Epicerie.' There is a printed Avis, signed T. Aubert, which 



