588 



SCIENCE 



[N. S. Vol. XLII. No. 1087 



traversing, nor can lie read the geograph- 

 ical evidence which lies patent to a trained 

 observer at every point of the journey; 

 much, therefore, of what he records may 

 be of interest, but probably lacks data 

 which are essential to the geographer if he 

 is to understand the geographical charac- 

 ter of the region and utilize it properly. 



Thus it happens that although the amount 

 of geographical material which is being 

 garnered may be large, the proportion of 

 it which is available for use in a scientific 

 investigation of an area is smaller than is 

 probably realized by those who have not 

 made the experiment. And yet it is only 

 by this scientific investigation of selected 

 localities or of a single phenomenon and 

 by working them out as thoroughly as pos- 

 sible that any real advance in geographical 

 science can be made. The accounts of such 

 pieces of work will not appeal to those who 

 desire picturesque descriptions of little- 

 known lands, but they will be welcomed by 

 geographers who can appreciate the value 

 of such studies. There should now be an 

 ever-increasing number of such geograph- 

 ers, trained to proceed in their investiga- 

 tions by the true scientific method, and 

 there should be a very considerable amount 

 of sound work in various branches of the 

 subject which aims at thoroughly investi- 

 gating some phenomenon, or group of phe- 

 nomena, so as to present a grouping of data, 

 carefully verified and critically discussed, 

 in order to arrive at conclusions which may 

 form a useful addition, however small, to 

 the sum of our geographical knowledge. 



So far as I am able to judge, the output 

 of serious work of this character is not 

 nearly as large as it should be, and I would 

 indicate some fields in which there is a lack 

 of individual work of this character. Until 

 more of it is undertaken we shall lack in 

 this country the material from which the 

 foundation of scientific geography can be 



built up, and while our own islands and 

 the various parts of the British Empire 

 furnish unrivalled opportunities for such 

 work, there are still far too many subjects 

 where the most thorough investigations 

 have been made in other countries. 



Mathematical geography presents a field 

 for research which had comparatively little 

 attention paid to it in this country. In 

 many respects this part of the subject is 

 peculiarly suitable for such treatment, since 

 it admits of the employment of precise 

 methods to an extent which is not always 

 practicable in cases where so many of the 

 factors can only be approximately defined. 

 The determination of positions on the 

 earth's surface is carried to great refine- 

 ment in the national surveys of most civil- 

 ized countries in order to furnish the neces- 

 sary controls for the preparation of large- 

 scale maps, but when we pass to the location 

 of travellers' routes, where considerable 

 allowance has to be made for the conditions 

 under which the observations have to be 

 taken, we find that very inadequate atten- 

 tion is usually paid to the discussion of the 

 results. Usually a mean value for each lati- 

 tude, longitude or azimuth is obtained by 

 the computer, and he remains satisfied with 

 this, so that when the route of another 

 traveller follows the same line or crosses it 

 at one or more points, it is almost impos- 

 sible for the cartographer to say which of 

 the two determinations of any position is 

 entitled to the greater confidence. In this 

 class of work, whether the results are ob- 

 tained from absolute observations at certain 

 points or from the direction of march, and 

 the distance traversed, it is quite practica- 

 ble to determine the range of uncertainty 

 within which the positions of different 

 points are laid down, and it is eminently 

 desirable that this should always be done in 

 order that the adjustment of various routes 

 which may intersect in partially-known 



