Livingston survey data from 1958 to 1971 are used to test the hypothesis that price expectations are normally distributed across the respondents to each survey. If the forecasts of the CPI and WPI are grouped around integers, chi-square tests indicate that normality is not a bad hypothesis. On balance, however, the sample distributions are skewed to the right and more peaked than the normal. Other distributions were therefore tried and even better fits obtained with a scaled logt-distribution.