A 'phase-shift' technique for detecting photometric companions to cepheids is applied to 202 variables. The inferred incidence of duplicity is 35 plus or minus 5%. An earlier 'color-color loop' method suggested 20%-27% duplicity. However, when positive results are required of both methods for a given star, the incidence falls to 15%. This lack of detailed agreement, as well as an apparent correlation of duplicity with period, suggests neither method is infallible.