432 SCIENCE PROGRESS. 



or two cases the information has been courteously supplied 

 in answer to special application, by directors of observatories 

 whose publications were not accessible, but it has mainly 

 been extracted, with the sanction of the Kew Committee 

 of the Royal Society, from the official publications received 

 by the Kew Observatory. Few of these publications can 

 be consulted except by those who have access to a good 

 library, and frequently mean annual values are not explicitly 

 given but have to be deduced by more or less arduous cal- 

 culations. The numerical measures of force are all given 

 here in terms of C. G. S. units, measurements in terms of 

 British units having been transformed when necessary. 



Taking a well-equipped magnetic observatory, we find 

 two distinct sets of observations in progress. There are, 

 first, the absolute observations, in which numerical values of 

 the declination, dip and horizontal force are obtained at 

 intervals from direct observations. There are, secondly, the 

 magnetic curves from the self-recording instruments, called 

 magnetographs, which supply a practically continuous record 

 of the variations in the values of the declination and the 

 horizontal and vertical forces. The curves by themselves 

 supply a large amount of qualitative and even to a certain 

 extent of quantitative information, but for the deduction of 

 a satisfactory value for such a quantity as an annual mean 

 they must be standardised by the aid of the absolute obser- 

 vations. 



In a magnetic curve distances in one direction, taken as 

 abscissae, measure the time, while distances in a perpen- 

 dicular direction, taken as ordinates, give the differences of 

 the magnetic element considered from a certain value 

 answering to the base line. The value of unit length of 

 the ordinate is obtained by direct experiments performed, 

 say once a year, on the magnetographs. The determina- 

 tion of the value of the element answering to the base line 

 is effected by measuring the ordinates corresponding in time 

 to the absolute observations, and for various reasons it is 

 expedient to make this comparison pretty frequently. 



When the magnetic curves have been standardised they 

 may be employed to get the diurnal oscillation and mean 



