How many pure math papers are published a year? I vaguely remember seeing a figure of 10,000 but that might be old, and I may be wrong.

6$\begingroup$ I think you're off by an order of magnitude. For instance, MathSciNet reports 111870 papers with a publication year of 2013. $\endgroup$ – Nate Eldredge Mar 18 '16 at 16:54

4$\begingroup$ @NateEldredge that's not only "pure math" though. $\endgroup$ – user9072 Mar 18 '16 at 17:15

$\begingroup$ Neither does it contain all "pure math" publications. $\endgroup$ – Franz Lemmermeyer May 15 '16 at 12:23
At SCImago you can find pretty much the entire statistics:
The first graph gives the total number of math papers per year. The second graph breaks it down per subject area (so you can distinguish "pure" from less pure). I would say that about 20% is "pure math", so some 30.000 papers per year. The number 10.000 you mention was the annual count twenty years ago.

1$\begingroup$ "Applied mathematics" has little meaning... most likely means some undetermined subject, probably often part of one of the various others (num. analysis, control and optimization, math. physics, th. comp. science, modeling and simulation, statistics and probability). I hope that not too much is deduced from such imprecise "statistics". $\endgroup$ – YCor Mar 4 '19 at 10:54