Editor's Note: Peter's original document lives at the Stardial Home Page. He suggested that this short item might make a good Show-and-Tell story, and I agree. I've reformatted his text slightly, and inserted the pictures into the text. MWR 8/18/98.
To demonstrate what would happen if your drift-scanned away from the celestial equator, PRM drift scanned a 512x512 CCD across the Andromeda Galaxy, M31 (declination +41.25 deg) one partly cloudy night using a mount that could slew eastward approximately 16 times faster than the sidereal rate.
He clocked the charge out at the best rate for the center of the CCD. That rate was necessarily incorrect for the northern and southern parts of the CCD. And because the stars' paths are arcs, rather than straight lines, even at the center of the CCD, the images of stars are distorted: at the center they look like tadpoles with their tails pointed northward. In an ideal case, the images of stars should only be a function of declination, but in this case, even that isn't true, presumably because the night was partly cloudy and thus the light from stars was modulated as the stars crossed the CCD.
There are at least three ways to remedy the unusual stellar images illustrated above:
Alternatively, aren't those tadpole-like stellar images artistic?
By the way, the capability of drift scanning 16 times faster than the sidereal rate could be useful for de-aliasing light curves of variable stars that are observed once per night by stationary cameras such as Stardial or TASS, amongst other uses.
Here's an image taken with the same technique but on the celestial equator (and near the Galactic plane). A periodic error in the drive mechanism is thought to cause the elongation in the vertical (scanning) direction. The square pixels are 25 arcseconds on a side in each case.