( 297 ) 



2. We think we can show that the writers have expressed them- 

 selves here in too definite terms, and that a distinctly positive result is 

 obtained if another method of research is followed. 



Already in furnishing the mean monthly temperatures for periods 

 of 5 years, which the writers used for their research, the present 

 writer expressed his doubt as to whether these data were lilted for 

 the purpose aimed at. If' the fluctuations of the mortality of infants 

 were merely directly proportional to the fluctuations of the mean 

 monthly temperature, the relation sought for oughl indeed to appear 

 also in this way. As soon however as these data are otherwise 

 connected e.g. the increased mortality occurs only after the tem- 

 perature exceeding a certain limit — it is no longer allowed to use 

 mean values without further inquiry. 



The research of Prof. Saltkt and Mr. Falkenbürg did indeed not, 

 as we saw before, reveal any relation between mean monthly tem- 

 peratures and mortality of infants. We will demonstrate this here 

 clearly onee more by giving for a single town, Groningen, the 

 deviations from the mean for 25 years of the 5-annual means of 

 the mortality of infants in the summer months May-September, ami 

 by their side, in italics, the same deviations for the temperature. 

 The former are given in hundredths per diem, the latter in tenths 

 of a degree Celsius. 



GRONINGEN L881— 1905. 



If high temperature-means for 5-year-periods were accompanied by 

 high mortalities of infants, at least the signs of the two kinds of 

 deviations ought to agree generally. We find however 12 concordances, 

 12 discrepancies and 1 undecided case; hence in this way no relation 

 is apparent. 



3. It might be supposed that this very unfavourable result is 

 caused partly by adding up the numbers for five years. In this con- 

 nection we might mention that Dr. Olland ') formerly traced a 



x ) A. G. Olland, Invloeil van het weer op het sterftecijfer. Thesis, Utrecht. 1890. 



