number of books exchanged in any one day, however, is not large, and in view of the rapid spread of the infection among the pupils, it is not felt that this is an important means among the possible agents of transmission.

A recess of about 10 minutes is permitted twice daily, and during this time there is free intermingling of the children, mostly in the open, however, weather permitting. The pupils who bring their dinners are not required to eat in their own seats, and they have a tendency to collect in groups about tables. No food is served at the school, no dishes are supplied, and the exchange of food is not common. There is, of course, abundant opportunity for "droplet infection," as well as more indirect methods of transferrence of saliva from person to person. The attack rates in the different schoolrooms were as follows:

In studying the chronology of the epidemic among different occupational groups it was found that, with the exception of the school as above noted, the other occupations were affected practically synchronously, and it would seem, therefore, that these may be considered as subjected to about the same opportunities of infection.

TABLE V.--Incidence of influenza in persons on Kelleys Island known to have attended social gatherings in January, 1920.

Other possible influences considered: Parties.-A number of parties were held upon the island just prior to and during the early part of the epidemic, as follows: January 18, 20, 24, 27, and 28. (See Table V.) It will be noted from Table V that these parties were

attended by from 7 to 30 people. It is difficult, however, to estimate the effect, if any, which these parties may have had in spreading the disease, owing to the fact that the number of people at any one party was small, the same people often attending several of them, and it is known that many persons in attendance had definite exposure to multiple possible sources of infection at school, in their own home, or elsewhere. The party held on January 29 deserves especial mention. This was a masquerade dance held in the village hall and attended by 30 persons. A buffet lunch with a beverage was served, the latter by means of individual paper cups. In attendance at this affair there were three people who were coming down with influenza and who were actually ill at the time. Six people were present, moreover, who had cases of influenza in their families at home. It would indeed seem probable that some cases might develop from such exposure, and the high attack rate in this group (76.8 per cent), notwithstanding the small number involved, may be significant. The school children in attendance were mainly from the high school, which had the highest incidence of any room in the school (71.1 per cent), and it may be that there is a relation between these two attack rates.

Milk as an agent of spread.-Fresh milk for the island is entirely a local product. The individual dairies are too small in most cases, however, to permit of separate study, hence we have grouped them for this purpose into those supplying a single household and those which produce an excess for sale to other families. It would seem that in the former group milk could not be a factor in introducing the infection into the home, although in the latter such a possibility might exist through the milk, returned containers, etc. Reference to the summary of this study, Table VI, shows, we believe, that milk was not a factor of importance in spreading the influenza at Kelleys Island. A study of the chronology in the different groups, moreover, leads to the same conclusion. (Table VII.) A further study of those families using milk from dairies which had influenza in the household failed to show significant variation either in attack rate or chronology when compared with users of milk from dairies having no ill attendants.

TABLE VI.-Incidence of influenza in families receiving their milk supply from various

TABLE VIII.-Incidence of influenza in families according to source of water supply,

Water as an agent of spread.-A summary showing the principal sources of water supply and the attack rate among users of each is given in Table VIII. Rain water is collected in family cisterns. No fresh water had been stored for over two months, owing to lack of rain. In many of the homes the supply of cistern water was low and was supplemented by water carried or hauled from Lake Erie. Others used raw lake water as their chief supply. Two drilled wells on the island supplied two families. No evidence incriminating any one of these sources as a carrier of influenza was found.

Insects. Insects need not be considered, as the weather was too cold for them to be active.

Crowding. The possible influence of crowding, as indicated by a consideration of the number of rooms per person in the various households, is shown in Table IX. In 1920 it will be noticed there is very little apparent relation between the number of rooms per person and the attack rate, whereas in 1918 there is a tendency for the attack rate to vary directly with the crowding. This question will be considered later.

Economic status.-There is not a wide range of economic conditions obtaining on the island; consequently this locality is not as suitable for studying this question as a region where the extremes of variation are more pronounced. The economic status was recorded entirely from the general appearances and conditions of the home. The results of this study are shown in Table X. It will be seen that in 1920 the attack rate was lowest in the lowest economic group, whereas in 1918 the reverse was the case.

Housing conditions.-A consideration of the attack rates among those living in poor, average, and good houses is shown in Table XI. It will be noticed that in 1920 the poorer houses were less visited by influenza, whereas in 1918 the reverse was the case.

TABLE IX.-Incidence of influenza in relation to rooms per person in dwellings, 1918

and 1920.

TABLE X.-Incidence of influenza in relation to economic status of families, 1918 and

TABLE XI.—Incidence of influenza in relation to housing conditions of families, 1918

and 1920.

TABLE XII.--Incidence of influenza in relation to general sanitation in the homes, 1918

and 1920.

General Sanitation.-A consideration of persons living under poor, average, and good general sanitation in their homes is shown in Table XII. It will be noticed that the apparent influence of this factor is also exerted in opposite directions in the two epidemics of 1918 and 1920, respectively. Generally, it may be said that those persons whom we usually consider the more fortunately situated suffered relatively more severely in the 1920 epidemic than did their poorer