84 G. H. KNIBBS. 



In these expressions D u D 2 , D_> denote the differences of 

 the daily averages for the quarters (with the proper signs): 

 Ti, T 2 , T 3 , the intervals, all being regarded as positive : and 

 Q the length of the uniform quarter in days. 



In the case of the correction for the half-year, the daily 

 average being still the datum, we have for the correction 



<»> • - ¥ 



D being the difference (with its proper sign) of the daily 

 averages for the two 'half-years; T the interval between 

 the end of June and the middle of the year, reckoned posi- 

 tive : and S the length in days of the uniform half-year. 

 The correction is positive for the first half year and nega- 

 tive for the second. 



These corrections are often so small that they may be 

 neglected. Corrections for the unequal lengths of the years 

 in a decennium are usually quite negligible. They may, of 

 course, be computed on the same principles. 



5. Equalised months and years.— Statistical results need- 

 ing analysis can also be put in a suitable form by using as 

 original data results for equalised months and years. The 

 scheme should be as follows: — 



Inasmuch as the year 2000 is a leap year, the period of 

 199 years lasting from 1st January 1901 to 31st December 

 2099 may be regarded, for statistical purposes, as consisting 

 of years of 365J days. Both months and years may then 

 be readily equalised by making the months 365J -f- 12 = 

 30-rs days, and considering the first of the month as falling 

 differently in each year of the cycle of four years, accord- 

 ding to the following scheme, viz.: — 



