Just a few points on this interesting discussion:
Philip doesn't "throw out" records but if he doesn't get a match on at least 2 out of 3 this will obviously require a manual intervention. TWYS may not be "wrong" but it will eliminate information about line feeds that will affect the interpretation of the data in De Long's logs, at least. This may not be serious if Philip doesn't distinguish between, say, "Cir Cum" and "Cir-Cum", but we don't know.
I am only saying that we could save him some work if he gave us a hint about how we should treat these compound cloud codes. The usual argument against this is that transcribers not visiting the forum won't know what we agree upon. However, I don't think this concern is very important for the Jeannette since, so far, those visiting the forum account for well over 90% of the transcriptions for this ship.
I am probably tiring you repeating all this so I will stop now.