IMPORTANCE OF VARIABILITY 93 



The reader cannot too sharply distinguish between the effect 

 of mean temperature and the effect of change of temperature. 

 The conditions seem to be closely analogous to those of baths. 

 The first plunge into a cold bath is highly stimulating. Almost 

 everyone is benefited. But what happens if the bath is prolonged.'' 

 Unless people take vigorous exercise there is a prompt reaction 

 which soon brings on a chill, and may end seriously. With a hot 

 bath the effect is the reverse. The first plunge brings on a sense 

 of lassitude, a desire to lie still and do nothing. So long as the 

 bath remains hot this feeling persists, and if the bath is prolonged 

 one finds one's will power declining. It becomes harder and 

 harder to summon up the decision to take a cold douche. Thus 

 it happens that even in winter a drop in the temperature of the 

 air is stimulating even to sick people. If the temperature remains 

 low, however, the good effect speedily passes and the deathrate 

 rises. 



In studying Figure 20 one is led to inquire whether the total 

 effect of falling temperature is more beneficial than the total 

 harm done by a rise. The answer is that the two are bound to be 

 equal. Otherwise the deathrate would keep declining until it fell 

 to zero, or would keep rising till everyone died. This sounds at 

 first as if the good and the harm of a variable climate balanced 

 each other, thus leaving people in the same condition that would 

 prevail if the temperature remained constant. This is by no 

 means the case, however, as appears from the following table, 

 based on the daily deaths in New York City from 1877 to 1884. 

 In preparing these figures we have first largely gotten rid of the 

 effect of daily changes by taking the average number of deaths 

 for overlapping periods of five days, that is, the first to the fifth 

 of the month, the second to the sixth, the third to the seventh, 

 and so forth. Generally such a five-day period contains some 

 days with a rise of temperature and some with a fall, so that the 

 two tend to balance each other. Then we have divided each month 

 into two halves, and from each half have selected the five successive 



