62 PROF. BALFOUR STEWART AND MR. W. DODGSON ON 



7. When the corrections of Table I. have been applied, 

 we obtain results indicating in minutes of arc the mean 

 diurnal declination-range, and which we may suppose to be 

 reasonably well freed from the influence of yearly variation 

 as well as from that due to locality. It is proper to remark 

 that the Prague ranges have been deduced from the obser- 

 vations taken at the hours 18, 22, 2, 10, as these hours 

 were common to the whole series with the exception of 

 1853. It seems possible that the result of confining our- 

 selves to these hours has been to make the ranges for 

 Prague alone somewhat too small. We do not, however, 

 regard our results as any thing more than a first approxi- 

 mation. It is likewise necessary to remark that for those 

 months during which we have two sets of observations we 

 have taken the mean of both. The monthly means are 

 recorded in Table II., those in brackets being interpolated 

 values. 



8. Our next operation has been to transform the 

 monthly values of Table II. into quarterly ones, four of 

 these going to each year. This is done by adding the 

 monthly values together in threes. We have not divided 

 the sums so obtained by three, preferring to adopt the 

 scale of 20", which we, of course, obtain by adding three 

 monthly values together without subsequent division. 

 This scale will therefore henceforth be adopted in this 

 communication. 



9. The next point is to arrange these quarterly values 

 into series of various lengths, and find for each series the 

 mean departure from the mean of the whole, as already 

 described in art. 2 of this paper. 



As we are now dealing with quarterly values, a series 

 arranged for eleven years will contain 44 terms, one for 

 twelve years 48 terms, and so on. In some few cases we 

 have deduced the mean departures by means of yearly 



