News came yesterday that Donald Keedwell has died.
Donald Keedwell was the author (with J. Dénes) of a classic, Latin Squares and their Applications, first published in 1974 with a foreword by Paul Erdős, and still the best source of reference. In my work with Rosemary Bailey, Cheryl Praeger and Csaba Schneider, we used a theorem of Frolov for which the best reference we could find was Dénes and Keedwell. The book is his lasting memorial.
Donald was a stalwart of British combinatorics. He organised the British Combinatorial Committee at the University of Surrey in 1991, and was a member of the committee for some time, holding the office of Secretary (if my memory is correct) for a while.
Peter Cameron's Blog 
This is sad and a great loss, I loved his book with Denes.
My only story is I asked him if it was really true if Sade of the man who enumerated all latin squares of order … (I forget off the top of my head)… was really related to Marquis de Sade and the answer was yes! This is obviously funny because as anyone who has tried to enumerate such things knows — this is sadistic act.
You can still see the pretty tribute to Keedwell that appeared on the 2008 2-day colloquium webpage
http://www.cdam.lse.ac.uk/colloquia-in-combinatorics-2008.html
I didn’t really know Donald very well; my interactions with him were limited to a couple of the LOOPS conferences and a couple of email exchanges in 2009 where he asked me some questions about loop theory.
Donald actually played a surprising role in one of my papers. He unknowingly found the first example of a Bol loop of odd prime power order with trivial center. Here is what Tuval Foguel and I wrote about it in our 2018 paper:
“While searching for particular types of projective planes, Keedwell gave an example of a normalized latin square of order 27 with certain properties. He published the example in two papers in 1963 and 1965, respectively. Denes and Keedwell later included the example in their well-known book on latin squares.
A normalized latin square is the Cayley table for a loop. It is obvious by inspection that the loop from Keedwell’s latin square has exponent 3, but no other properties are immediately apparent. Perhaps because the example appeared in the latin square literature, the loop theory community seems to have been unaware of it.
Using the LOOPS package for GAP, we analyzed Keedwell’s loop and found that it was, in fact, a left Bol loop with trivial center….”