4 
HARKER : PRESIDENTIAL ADDRESS 
are variations, both seasonal and casual, we must take a sufficient 
number of observations, properly distributed in time, and an average, 
duly weighted, will then give us as good a result as the nature of the 
case admits. But now suppose that we wish to know the amount of 
sediment carried by all the rivers of the world. We have data for 
nine rivers, data which are likely to differ much in respect of probable 
error. Accepting them, however, as they stand, it appears that the 
water of the Rio Grande carries one part in 291 of sediment, that of 
the Uruguay River only one in 10,000, the other seven rivers giving 
intermediate values. The highest figure is thus 34 times as great as 
the lowest. Some geologists will simply take a mean of the nine 
figures and proceed contentedly to use this result in the most far- 
reaching conclusions. I do not believe that a mean of nine figures so 
discordant can afford any information of quantitative value. The 
average must be extended over a much wider area, before a result is 
obtained of which we can legitimately make use. 
Where dynamical principles enter into the problem, the pitfalls 
w^hich aw^ait the unwary are sometimes less evident. I will take as an 
illustration the case of models, such as have been constructed to 
elucidate the mechanism of folding and faulting. In no case, so far as 
I am aware, have geologists had regard to the conditions which are 
necessary in order that a model may correctly represent the working 
of the original. The various forces concerned must bear their proper 
ratios. Since the weight (for a given material) is reduced propor- 
tionally to the cube of the linear dimensions, the other forces must 
be reduced in the same ratio, and it is in fact impossible to make this 
adjustment as regards the internal forces which resist deformation 
and fracture. Moreover, the velocities of the moving parts should 
be reduced in proportion to the square root of the linear dimensions ; 
and this makes it hopeless to think of imitating the slow processes 
of mountain-building. Models of this kind may afford useful geo- 
metrical illustrations, but can throw no light on dynamical problems. 
The same remark applies to models of glaciers, but here there is no 
need to go to artificial models to illustrate my point. Some geologists 
still argue from the behaviour of an Alpine valley-glacier to that of a 
continental ice-sheet, without perceiving how completely the different 
scale of magnitude must modify the mechanical conditions. 
Experiment has undoubtedly afforded valuable help in the study 
of particular questions in the domain of physical geology, and this is to 
be recognized with gratitude. As regards the larger and more complex 
