A survey of this type cannot be used to point to any definite factor or factors predisposing to the development of hallux valgus. Nevertheless, a comparison of measurements in the morbid and control groups shows several outstanding differences:

1) There was a high degree of correlation between valgus and intermetatarsal angle in the two groups combined (coefficient, 0·7) but the correlation was higher in those cases with a degree of valgus greater than 25 degrees than in the remainder (coefficients, 0·36 and 0·53).

2) In the control group the first metatarsal was longer than tile second by a mean measure of 2 millimetres; in the morbid group by a mean measure of 4 millimetres. For a high degree of valgus and a low intermetatarsal angle the first metatarsal tends to be longer than the second by a significantly greater amount than when the high valgus is associated with iligh intermetatarsal angle.

3) In 90 per cent of the control cases there was a lateral displacement of the medial sesamoid of the first metatarsal of 3 degrees or less, whereas 88 per cent of the morbid group showed a displacement of 4 degrees or more. There was very little overlap in the distributions of this observation in the two groups. There was a high correlation between the degree of this displacement and the severity of hallux valgus.

4) Rotation of the hallux was not observed among the controls; in the morbid group those cases showing rotation had an average degree of valgus of 36 degrees while the rest had an average of 19 degrees. The mean degree of valgus in the morbid group was 32·0 degrees and that of the controls 15·5 degrees. The mean angle between the axes of the first and second metatarsals was 13·0 degrees in the morbid group and 8·5 degrees in the controls.

Since tile morbid group consisted largely of women (98 per cent) it is important to know that in the control group the only measure showing a statistically significant sex difference is that of intermetatarsal angle, but that, even so, the mean difference is only 1·3 degrees. Thus tile sex difference between the two groups is probably only of minor importance. The role of age in influencing the observations cannot be clearly elucidated from the data at present available. It can only be stated that there is no positive indication that age is a controlling factor in the departure observed in the morbid group from the control observations.