in the same occupation at the census of 1900. We find, for instance, that among agricultural laborers in New Jersey there was 1 divorce recorded to every 113 married males, among farmers 1 to every 92 married males, among actors and professional showmen 1 to every 6 married males. These ratios are not to be regarded as divorce rates. For instance, the ratio for actors does not mean that in each year one-sixth of the married actors procure a divorce; for the comparison is between the total number of actors obtaining a divorce in twenty years and number of married persons in this profession on a given date,-namely, the date of the census of 1900. A nearer approach to a divorce rate would be obtained if we took one-twentieth of the total number of divorces recorded, which would be the annual average, and compared it with the number of married males in the profession. That would make the annual divorce rate for actors 1 to 120 married males. But not much reliance could be placed upon a rate obtained in that way and from these incomplete data. I am of the opinion, however, that these ratios have value for comparative purposes. They do indicate or, at least, suggest that divorces are much more frequent among actors than among farmers and rather more frequent among farmers than among farm laborers.

Since the Bulletin was published, I have extended this kind of a comparison to include eight other states, and have ranked the occupations in each state with reference to the ratio of divorces to married males. The states included are Rhode Island, New York, Pennsylvania, West Virginia, Indiana, Illinois, Michigan, and South Dakota. In Rhode Island the statistics of occupations cover 44 per cent. of the total number of divorce cases; in New York, 37.3 per cent.; in Pennsylvania, 52.4 per cent.; in West Virginia, 36.7 per cent.; in Indiana, 33.6 per cent.; in Illinois, 33.9 per cent.; in Michigan, 45.6 per cent.; in South Dakota, 50.8 per cent.

Thus it appears that even in these selected states the returns are far from complete. If it could be assumed that the deficiency was approximately the same in all occupations, the

figures would be as satisfactory as complete returns. But of course no such assumption can safely be made. At the same time, as regards the state of New Jersey, where the distribution by occupation of husband is shown for 81 per cent. of the total number of divorce cases, it seems hardly probable that the corresponding distribution of the remaining 19 per cent. would differ so radically as to destroy the value of the comparison which has been made; and the results for other states are, on the whole, fairly consistent with those for New Jersey.

Assuming, then, that the limitations and defects of the returns are not so serious as to destroy their statistical value for the kind of comparison which I have here made, and in which the emphasis is laid, not upon the ratio, but upon the rank of the several occupations as determined by these ratios, it becomes interesting to study the results.

The popular impression regarding the frequency of divorce among actors here finds confirmation. In the ratio of divorces to married males this occupation or profession leads the list in all but one of the nine states for which the ratios referred to have been computed. (See Table IV.) In that one state (South Dakota) musicians and teachers of music have the first rank, and actors share the second rank with commercial travellers, physicians, and telephone and telegraph operators. It frequently happens that in a given state two or more occupations have the same ratio, and therefore the same rank, although the exact ratio expressed as a fraction or carried out to a sufficient number of decimal places would doubtless show differences. But such differences have not been computed. Musicians and teachers of music, as just stated, have the first rank in South Dakota. In one other state they share the first rank with actors. They rank second in three states, third in another state, and fourth in the one remaining state. On the whole, this occupation appears to be the one which most closely rivals that of actors in the divorce courts.

Commercial travellers have the second divorce rank in four states. In four other states their rank ranges from third to sixth, but in the remaining state, Indiana, they have, for some

reason, an exceptionally low rank, being thirtieth. Perhaps some of you may feel that I ought to say exceptionally high rank, but of course the terms high and low are here used in a purely statistical sense, and imply no moral judgments.

Barbers and hairdressers represent another occupation in which divorce appears to be relatively frequent, the rank of this occupation ranging from third in the state of West Virginia to eleventh in the state of South Dakota.

Bar-tenders have comparatively high divorce rank in six of the nine states, a medium rank in two other states, and in the remaining state, Rhode Island, an exceptionally low rank, being, in fact, thirty-seventh. It is not improbable that an exceptional ranking appearing for a single state reflects some inconsistency in the occupational classification resulting from peculiar or local usages in regard to designation of the occupation. For instance, a bar-tender might be returned and classified as a saloon-keeper.

In most of the states hotel-keepers are well up towards the head of the list as regards divorce, and so are restaurant and saloon keepers.

Having specified those occupational classes which appear to be most addicted to divorce, let us now inquire what occupations represent the other extreme, having what I have rather hesitatingly termed a low rank in this respect. The comparison we are considering distinguishes thirty-nine occupations or occupation groups. In four states, Rhode Island, New York, West Virginia, and Indiana, the thirty-ninth or last place on the list is occupied by farm laborers. In Pennsylvania and Michigan this distinction belongs to miners and quarrymen. In New Jersey stationary engineers and firemen come at the foot of the list. In Illinois clergymen have this position; in South Dakota, farmers.

The range in the ranking of farm laborers is from twenty-third to thirty-ninth in the different states included in our comparison. For blacksmiths the range is from twenty-fifth to thirty-seventh; for carpenters, from twenty-fifth to thirty-sixth; for farmers, from twenty-third to thirty-ninth; for clergymen, from thirty

third to thirty-ninth. Next to farm laborers, clergymen appear to have the lowest average rank as regards divorce.

The fact just stated regarding the clergy suggests a comparison with the other so-called liberal professions, law, medicine, and teaching. I have already referred to the high rank in the divorce record indicated for physicians. The range extends from the second rank in South Dakota to the eleventh in West Virginia. It is not altogether clear whether the second place among the professions belongs to lawyers or to teachers. Each of the two professions show a rather wide variation in ranking. For lawyers the range extends from the fourteenth rank to the thirtyfifth, for teachers from the sixteenth to the thirty-second. On the whole, however, there appears less divorce among teachers than among lawyers. In two states, New York and Indiana, lawyers and teachers seem to be in the same class with clergymen; but in the other states their divorce rank is considerably higher.

The tables which follow present the data upon which the above discussion relative to occupations is based. Table I gives the number of divorces granted during the twenty years 1887 to 1906, classified according to the occupation of the divorced husband; Table II, the number of married males reported in each of the specified occupations at the census of 1900; Table III, the ratio of divorced husbands to married males in each occupation; and Table IV, the rank of the several occupations as determined by this ratio.

In these tables the attempt has been made to arrange the occupations in the order of the prevailing or most usual rank shown for the nine states included in this comparison. This has been determined partly with reference to the median rank and partly with reference to the portion of the scale where there appeared to be the greatest concentration in the ranking of the given occupation. The order, which was thus determined without applying any mathematical formula, differs slightly from an order determined by the average rank, but is believed to be more typical and representative.