436 SCIENCE PROGRESS. 



to the discovery of a scale value, but this cannot be trusted 

 to for absolute results in view of the change that has pro- 

 bably occurred. In such cases it seems preferable to em- 

 ploy the records of the absolute observations, and in some 

 observatories these are the only results obtainable. Abso- 

 lute observations, if a sufficient number exist, are satisfac- 

 tory enough, when corrected for the diurnal inequality. 

 In some instances where no such correction seems to have 

 been applied, the hour or hours of observation may have 

 been chosen so as to eliminate the inequality. Unfortun- 

 ately information on this point is commonly deficient, and it 

 might be as well in such cases to regard the results as 

 uncertain to the extent of at least half the daily range. 



This uncertainty is in reality not so serious a matter as 

 might appear at first sight, for observatories as a rule are so 

 conservative in their instincts that their results for successive 

 years, though possibly in strictness not mean annual values, 

 are fairly comparable amongst themselves. The uncertainty 

 arising- from possible differences between the instruments at 

 different observatories has also to be kept in view. An 

 interesting example bearing- on this point is furnished by 

 the Greenwich results in the table for dip needles of 

 different lengths. 



In the publications of most magnetic observatories force 

 is measured to five significant figures, and declinations and 

 dips to the nearest tenth minute, and when this is done the 

 practice has been followed in the present table. After the 

 warning given above as to probable differences between the 

 instruments at different observatories this can do no harm, 

 and it may prove useful in tracing secular variation if 

 similar tables be published in future years. 



The interval elapsing before the publication of results 

 varies much in different observatories, so that there is unfor- 

 tunately a considerable variety in the dates in the table. 

 Following the example of the B. A. Report for 1887, pp. 

 328-331 — a source from which a good many of the geograph- 

 ical data are taken — I have arranged the observatories 

 according to their latitude. The longitude is in all cases 

 referred to Greenwich. In some few cases where the geo- 



