Apeil 13, 1917] 



SCIENCE 



351 



This volume is so important to geodesists and 

 geologists that the undersigned believes that 

 its appearance warrants more attention than 

 merely a review of the ordinary kind and 

 length. Hence the present article is pre- 

 sented. The volume is of especial interest 

 because it is based on many more observa- 

 tions of gravity than were formerly available 

 in the Coast and Geodetic Survey, and be- 

 cause the discussion of the data is much more 

 searching and complete than anything here- 

 tofore published. 



The last preceding publication on this sub- 

 ject from the Coast and Geodetic Survey was 

 issued in 1912.^ The conclusions in that 

 publication were based almost entirely upon 

 the 124 gravity stations in the United States 

 and 10 in Alaska. In the present publication 

 conclusions are based primarily upon observa- 

 tions at 219 stations in the United States and 

 are tested by observations at 42 stations in 

 Canada, 73 in India and 40 in other parts of 

 the world — 374 stations in all. This increase 

 in the number of stations and of the area 

 covered by them is important because the 

 reliability of the conclusions drawn depends 

 upon the completeness with which effects 

 peculiar to a given station or a locality are 

 eliminated, and upon the extent to which sup- 

 posed general laws are found to be confirmed 

 by the evidence from widely separated areas. 



Within the past four years there has been 

 one radical change in the relative method of 

 observing gravity in the United States, with 

 the half-second pendulums, the standard 

 method of the Coast and Geodetic Survey. 

 Formerly each gravity observer carried a tran- 

 sit instrument with him and determined the 

 errors of his chronometers on each clear night. 

 Now he determines the chronometer errors 

 by the time signals sent out from the ISTaval 

 Observatory at Washington, special studies 

 having shown that these signals are of ample 

 accuracy for the purpose. The time and ex- 

 pense necessary for securing a gravity deter- 

 mination at a station is thus materially re- 



1 "Effect of Topography and Isostatio Compen- 

 satiou upon the Intensity of Gravity," by William 

 Bowie, Special Publication No. 12. 



duced, and the rapid increase in the number 

 of gravity stations which is the prime essen- 

 tial of future progress is facilitated. 



The most severe test which may be applied 

 to a theory, or a supposed understanding of 

 a group of facts, is a test by prediction. Such 

 a test has been applied by the writer recently 

 to the Coast and Geodetic Survey theories as 

 to the intensity of gravity, including the 

 method there used for computing the effects 

 of topography and isostatic compensation. In 

 the 1912 gravity publication already referred 

 to (Special Publication No. 12) there is a 

 map of the "lines of equal anomaly for the 

 new method of reduction." This map, based 

 upon 124 gravity stations in the United States, 

 may be used in combination with computa- 

 tions for which the methods have been fully 

 published, to predict the intensity of gravity 

 at any point in the United States. Such pre- 

 dictions, for 95 gravity stations, based on the 

 1912 map have been made by the writer,- and 

 compared with the observations made after 

 1912 at those stations. The difference of 

 prediction and observation was less than .02 

 dyne (1/49,000 part) in 54 out of the 95 cases. 

 Such a prediction may be made in about three 

 days by an expert computer in Washington at 

 a cost to the government of about $25. On 

 the other hand if a physicist undertakes to 

 determine gravity in his own laboratory by an 

 absolute method, past experience shows that 

 after many weeks of time and an expenditure, 

 including salaries, of many hundreds of dol- 

 lars, it would be an even chance whether his 

 final result would be within .03 dyne of the 

 truth. To reduce the probable error suffi- 

 ciently to be equal to that of the prediction 

 which may be made in Washington in three 

 days by a computer would require the equiv- 

 alent of months of time by an expert physicist. 



Still more accurate predictions are certainly 

 now possible by the use of the 1917 map, based 

 on 261 stations in the United States and 

 Canada, which is necessarily a decided im- 

 provement for prediction purposes over the 



2 The writer took advantage in making these pre- 

 dictions of detailed computations which had been 

 made at Washington of the effects of topography 

 and isostatic compensation at these stations. 



