1. Successive portions of boiled and mashed potatoes, roast pork, cooked, freeze-dried peas and ice-cream were taken in order to determine the total (sampling and experimental) error involved in the ‘duplicate analysis’ method of dietary survey.
2. These samples were analysed for water, nitrogen, fat, iron and energy.
3. The experimental error of the methods was also studied separately.
4. The sampling and experimental errors ranged from 0.3% for water in potatoes, to 23% for Fe in ice-cream. Sampling error was significantly greater than experimental error for all nutrients studied.
5. It is concluded that in reporting the results from dietary surveys, it is necessary to take account of the errors inherent even in the most precise methods.