[Retros] Long PG's ending in mate

[Elkies, Noam]

Hello all,

Mark Watkins (whose name you may recognize from
his work on Losing Chess) recently posed this question
on Chess Stackexchange:
https://chess.stackexchange.com/questions/29362

Instead of an arbitrary bonus for ending in checkmate,
it's more naturally two questions: the well-known length task
for a PG with a unique solution, and the less well-known
length task for such a PG that moreover ends in checkmate.
Mark now asks me in e-mail (and accepted my offer to
ask his questions on this retros list):

1) Is Caillaud's capture-free https://pdb.dieschwalbe.de/P0002187
(R1k2KNR/1npp1Qpp/p3pp2/1p3q1b/1P4PB/r1b2n1N/P1PPPP1P/r4B2
PG 38.5, #4495 in _Die Schwalbe_ 83 10/1983, 1. Pr.) also
the current length record for a sound proof game ending in checkmate?

2) What is the source of the 36.5 move proof game
2Q5/K3pp2/RpkpB1p1/B4PPr/2PrPPq1/P1PR1n2/1p1PNnb1/8  ?
Mark found it in his files and can neither reconstruct the attribution
nor locate it on PDB or elsewhere.

Thank you,
--Noam D. Elkies