IMPORTANCE OF VARIABILITY 89 



The years 1877 to 1884 were chosen for investigation because, 

 extraordinary as it may seem, those eight years with the four that 

 followed them are the only ones for which accurate daily records 

 of deaths appear to be available in any American city. At 

 present not even the weekly records in most places are accurate. 

 Instead of tabulating and publishing the actual deaths during 

 a given week or month, the reported deaths are given. In New 

 York City, for example, the physicians who send in the reports are 

 allowed by law an interval of thirty-six hours between the hour of 

 death and the hour of filing the reports. In practice this is 

 extended to three or four days. No record of daily deaths is kept, 

 and the deaths which are published as those of a given week really 

 belong in part to that week and in part to the last half of the 

 preceding week. It is much to be regretted that exact records of 

 the deaths each day are not available for all parts of the country, 

 for, as we shall soon see, such records are of the highest impor- 

 tance. With' the counting machines that are now used in large 

 cities it would take a clerk only three or four days to get out a 

 year's daily record even in New York. 



Some of the results of a study of the daily deaths m New York 

 City from 1877 to 1884 appear in Figure 20. The figures at the 

 top indicate the number of degrees by which the temperature of 

 a given day differed from that o-f the day before. Thus the left- 

 hand side of the diagram pertains to days when there was a sharp 

 drop of temperature and the right side to those when warm waves 

 occurred. The figures on the right and left margins indicate the 

 amount by which the deaths on a given day exceeded or fell short 

 of those occurring on the previous day. The figures have been 

 inverted so that good conditions of health may be represented by 

 high positions and bad health by low. Since the average number 

 of deaths per day was not far from 100, the figures in the margins 

 are almost equivalent to percentages. The four dotted lines rep- 

 resent the conditions during the four seasons, beginning with the 



