How to compute the average, variance, covariance, and correlation of two series in a recursive way. That is, starting with the statistic for N items, how does it change when item N+1 is added?
Google scholar (which is great, of course) ascribes a well-known article by Peter Neary to the unknown scholars P. Draft and C. Welcome. I wonder who they are?
Quite productive fellows, it seems.
And then the thesis was done. (2 Mb PDF link).