Thanks for the corrections, Tom and Mac128. After posting, I recalled from the "Click of Death" stories that Zip drives have a head retract mechanism. This article confirms Tom's statement that the Zip isn't Bernoulli:

http://www.byte.com/art/9508/sec11/art9.htm

Unfortunately, we are still stuck with the riddle about when it is best to sleep a Zip drive. Disk head land off and take off is certain to cause media wear, but so does constant access. If you use your Zip cartridge as a Mac system disk, inevitably some cylinders/sectors will be accessed more frequently. Does the Iomega algorithm for disk access versus wear account for use as a system disk?