As in "Before testing or reconfiguring, always mount a scratch monkey", a proverb used to advise caution when dealing with irreplaceable data or devices. Used to refer to any scratch volume hooked to a computer during any risky operation as a replacement for some precious resource or data that might otherwise get trashed.
This term preserves the memory of Mabel, the Swimming Wonder Monkey, star of a biological research program at the University of Toronto. Mabel was not (so the legend goes) your ordinary monkey; the university had spent years teaching her how to swim, breathing through a regulator, in order to study the effects of different gas mixtures on her physiology. Mabel suffered an untimely demise one day when a DEC engineer troubleshooting a crash on the program's VAX inadvertently interfered with some custom hardware that was wired to Mabel.
It is reported that, after calming down an understandably irate customer sufficiently to ascertain the facts of the matter, a DEC troubleshooter called up the field circus manager responsible and asked him sweetly, "Can you swim?"
Not all the consequences to humans were so amusing; the sysop of the machine in question was nearly thrown in jail at the behest of certain clueless droids at the local "humane" society. The moral is clear: When in doubt, always mount a scratch monkey.
ESR notes: There is a version of this story, complete with reported dialogue between one of the project people and DEC field service, that has been circulating on Internet since 1986. It is hilarious and mythic, but gets some facts wrong. For example, it reports the machine as a PDP-11 and alleges that Mabel's demise occurred when DEC PMed the machine. Earlier versions of this entry were based on that story; this one has been corrected from an interview with the hapless sysop.
A corespondent adds: The details you give are somewhat consistent with the version I recall from the Digital "War Stories" notesfile, but the name "Mabel" and the swimming bit were not mentioned, IIRC. Also, there's a very detailed account that claims that three monkies died in the incident, not just one. I believe Eric Postpischil wrote the original story at DEC, so his coming back with a different version leads me to wonder whether there ever was a real Scratch Monkey incident.