“Doubtful” applies to those cases in which the evidence although suggestive of influenza, is not complete enough to warrant such a diagnosis. The doubtful feature may be in the lack of too great a number of the symptoms enumerated, or the presence of symptoms which might be due to some other disease. Certain cases occurring at the same time with other cases of typical influenza in the same household, and which would otherwise have been recorded as doubtful, have been marked either “Probable” or “Yes.” _Standards of severity._—A purely arbitrary classification of severity has been adopted. Probably no two observers would agree exactly on a classification of this nature, but for the purposes of this study the following will suit all requirements provided the standard used is carried in mind throughout the comparison. If a patient with influenza is under medical care, and the case is one of ordinary severity, the usual period in which the individual is advised to remain in bed is one week. This is the basis of the criteria of severity. _Mild._—A case is recorded as mild if the individual has remained in bed three days or less; _Average_, if in bed four to seven days; _Severe_, if in bed over seven days. _Pneumonia._ This designation is added to that of “severe” only in case the physician made such a diagnosis, or if the evidence under “symptomatology” leaves no doubt as to the condition. Examples of individual exceptions to the preceding general classification are as follows: An individual in bed two days, but sick for three weeks might be recorded as average. A mother, with a family of sick children and who spent no time in bed may have been a severe case of influenza. In fact, we have allowed ourselves a certain latitude in individual cases in classifying both the diagnosis and the severity of the disease. In the final tabulation we have included both the “Yes” and the “Probable” as being cases of influenza. This has been done after a careful comparison of both groups. As a check upon the reliability of the work we have compared our results for the 1918 epidemic with those reported by Frost and Sydenstricker and have discovered that with regard to the general subject discussed in both studies there is close agreement. This is important in view of the long period that has elapsed between the first pandemic and the time of our survey, and because we are unable to compare our tables of incidence for 1918 with those for the city or the state at large. Our own records do not place the date of occurrence of the disease in 1918 any more closely than by month. We have compared our 1920 incidence curves with those of Massachusetts and find a close correspondence, particularly in the date of onset, peak, and disappearance of the epidemic. We have done likewise for the occurrence of the disease in the city of Boston at large (Chart XIV).