XIX COMMUNICABILITY OF PHTHISIS 393 The last line of the table gives the actual number of wives dying in each interval, obtained by substituting for 1000, the number found in the previous process. viz. 4,363. Exactly the same results would of course be obtained by calculating from the wives to the husbands. If there be N. medical practitioners in England and Wales, and n of these send in returns, then (assuming no infection, and assuming that those who send in returns have an average share of practice) they should have observed during the ten years a number of cases equal to 7 of the above numbers, though presumably they would only record a part of the cases observed. According to the Medical Directory for 1883, N = 15,462. But within the decade both husband and wife would die of phthisis, within a twelvemonth of each other, in 829 instances. If, therefore, 500 medical practitioners sent in returns as to their experience, they should have observed collectively about 27 such cases. This is the * probable ” number that would have been observed, but quite apart from infection the number actually observed might be greater, since the number of practitioners and number of years are not very large, and the coincidence is a somewhat uncommon one. Again, marriage may have some effect on phthisical mortality otherwise than through infection (e. g. sexual selection might be imagined to have a notable effect). It is plain, therefore, that to show any substantial argument for the existence of infection it would require a much larger collection of cases than has vet been published. 1 This number does not include army and navy medical officers, but on the other hand it does include many retired medieal men.