Phase-shifting interferometry is the standard method for testing figure err
or on optical surfaces. Instruments measuring spheres and flats are readily
available, but the accurate measurement of aspheres requires null correcti
on. One problem with the general (nonull) testing of aspheres is the loss o
f common path. Systematic errors are introduced into the measurement by the
fringe imaging optics. The sources and types of error are reviewed, as wel
l as their effect on a wave-front measurement. These nonnull errors are pre
dicted generally, with third-order analytic expressions derived for a tilte
d or a defocused test surface. An interferometer is built to test the expre
ssions. The imaging system is a single lens, nominally image telecentric. M
easurements are performed on a test surface defocused from -5 to 5 mm. The
resulting measurement bias is shown to be in goad agreement with third-orde
r aberration theory predictions. (C) 2000 Optical Society of America.