68 OLDHAM: THE STRUCTURE OF THE HIMALAYAS, ETC. 



Comparing the calculated with the observed deflections, it will 

 be noticed that either supposition gives results which are in very 

 fair agreement with reality ; only in the stations from Masi to 

 Nimkar is there any considerable irregularity, but these stations 

 are situated in the tract where northerly are passing into southerly 

 deflections, and where a small variation in the assumed form of 

 the trough would lead to considerable changes in the calculated 

 deflections. Apart from this, the general course of the variation, 

 as well as the actual values, of the calculated and the observed 

 deflections are in very good agreement ; at the northern stations 

 the calculated deflections are in slight defect, and the same is true 

 of the stations in the southern half of the section, but the former 

 of these is easily accounted for by the probable excess of the 

 northerly attraction of the Himalayas over that allowed for in 

 the calculations, or both the deficiencies could be eliminated by 

 assuming a rather greater depth of the trough, but no real benefit 

 would accrue from any attempt at obtaining a closer agreement 

 between calculation and observation. 



This study of one of the groups of latitude stations serves to 

 illustrate at once the method which will be followed, and the limit- 

 ations of any attempt to derive geological information from geo- 

 detic observations. The method, though differing in form, is 

 essentially the same as that adopted in geodetic work ; a certain 

 assumption is made, calculation is made on that basis and the 

 results of calculation and observation compared, another assump- 

 tion is then made and fresh calculations made until the average 

 difference between the calculated and observed values of the 

 deflection is reduced to the smallest amount. In geodetic work 

 proper the closeness of agreement is tested by comparing the sum 

 of the squares of the individual differences, and adopting the sup- 

 position which gives the smallest value to this sum, as the one 

 which most closely approaches the average conditions. This method 

 is the only oneadmissible where a large number of observations, 

 extending over a large area, have to be dealt with ; it is not only 

 unnecessary, but would give a wholly illusory appearance of pre- 

 cision, if applied to a limited number of observations, and to the 

 extraction of the information for which we are in quest. 



Here we must start with those conditions which represent a 

 near approach to the average, and apply to them a correction for 



6 216 ] 



