( 299 ) 



lalitv mus! be obtained if the latter are compared with mean tem- 

 peratures or numbers of days with high (einperatures for a period 

 which begins and ends a little earlier. 



We have tested this conclusion by calculating the mean tempera- 

 tures as well as the numbers of davs with temperature above 25° — 

 which, in accordance with a terminology used in meteorology, we 

 will call henceforth "summer davs" - once for the calendar months, 

 once for periods from the 16' 1 ' of one month till the 15 th of the 

 next month. 



Both the choice of the "shift" of J."ï days and that of the tem- 

 perature-limit 25 are somewhat arbitrary. For a preliminary research 

 like ours there is however no objection to this. From the table below 

 it appears that at least for some parts of our country the shift has 

 about the size which serves the purpose ; by S are indicated the 

 mean values of the mortality of infants for calendarmonths during 

 the period mentioned in the first column, given in percentages of 

 one twelfth of the yearly death-rale, as we will continue doing in 

 the following panes; by T the mean numbers of summerdays in 5 

 rears, for periods of one month shifted over 15 days. 



For Groningen and Utrecht the maxima now coincide wholly or 

 almost so. For the province of Zealand, only partially well represented 

 as to its climate by Flushing, a difference of a month in the epoch 

 of the maximum remains. 



4. We first give here the results of the investigation for Groningen, 

 in which we compare successively with the deviation of the mortality 

 in 5-year periods the deviations of: 



