an inevitable quantitative in­feriority easily looks like a qualitative weakness

if you torture the data enough, nature will always confess

I rarely comment explicitly on the merits of an article, but ‘EV 4, 1234’ indicates a piece less good than ‘EV 4, 1234 (Pecorino)’

tardi ingeni est rivulos consectari, fontes rerum non videre

An employee cannot have too much information

I had no idea of industry and diligence till I saw Bekker’s collations. There is scarcely an author in Greek who he has not collated with MSS

Younger sci­en­tists in par­tic­u­lar lose sight of the dif­fer­ence be­tween re­al­ity and sim­u­la­tion … They have grown up with the mod­els

analytic induction ends where enumerative induction begins; and if well conducted, leaves no real and soluble problems for the latter

No matter how early one starts, and how much work is put in, pitchbooks nearly always involve all-nighters just before the meeting

not strikingly bad on each occasion, but cumulatively spoils a writer’s style