( 302 ) 

 Hence so far as the signs are concerned we find : 



Concordance Discrepancy Undecided 

 Utrecht: 13 7 



Zealand: 12 4 4 



The number of concordances in these eases is large enough in 

 show that the fluctuations in the number of summerdays play an 

 important part in the matter of the mortality of infants, especially 

 if we consider that at Utrecht as well as in Zealand most concord- 

 ances occur in the months of large mortality of infants: July and 

 August for Utrecht, August and September for Zealand. Doubtless 

 there are besides them other important circumstances, not connected 

 with temperature, the influence of which even in means for 5 years 

 is not yet eliminated. This need not surprise us — it rather may 

 be called remarkable that in the case of Groningen not more is 

 shown of such circumstances. Whether a better agreement might lie 

 obtained by assuming a higher temperature limit for Utrecht, for 

 Zealand a lower temperature limit and another value for the shift 

 we have not tried so far. 



The mean temperatures for (lie periods May Hi" 1 — .lime J 5 th etc. 

 were not calculated for Utrecht — de Hilt and Flushing; a comparison 

 of the methods a, c and (/ however favours the view, that no impro- 

 vement would ensue. For, the result for Utrecht was : 



It would seem as if the result for Utrecht is practically the same 

 in the three cases. If however also the quantitative agreement is 

 considered, this appears to be wholly lacking in the case c, but 

 to be much belter in case </ than in case a ; hence the results obtained 

 for Groningen are also here, though not so generally, valid. 



5. A drawback of the use of means or sums for [periods of 5 



