G. W. Robinson 343 



and terrain, neither of the last two systems is alone suitable. This 

 is the case in the United States where a physical system of classification 

 is used. Lastly, there is the system of classification on ecological and 

 agricultural lines recommended by Hall and RusselU. This last system 

 has been adopted as far as possible by the writer in North Wales. Its 

 use is limited on the agricultural side by the operation of factors other 

 than those connected with soil and climate. On the ecological side 

 certain broad classifications can be made. Sand dune, moor, salt 

 marsh and other formations can be distinguished in waste lands, but 

 it is a matter of considerable difficulty to refer all cultivated lands to 

 their original plant formations. 



In classifying the soils of North Wales, genetic factors as well as 

 the intrinsic properties of soils have been taken into consideration in 

 addition to agricultural and ecological characters. By this means a 

 provisional classification has been arrived at which may be modified 

 by sub-division as more information is obtained. The system adopted 

 will be best explained by the accounts of the various soil types. 



The analytical methods adopted are those generally used in soil 

 surveys in this country^. 



Table I shows the average percentage composition of each soil 

 type. In order to give some idea of the dispersion of the results, 

 the range is given corresponding to each average. Thus in the case of 

 the Anglesey medium loam we see that the percentage of fine sand in 

 the soil is 21-2 and the range 16-1 to 29-2. This means that in all the 

 samples examined the percentage of fine sand was never less than 16-1, 

 nor greater than 29-2. The range is of course a rather imperfect 

 measure of dispersion but is the only possible measure where the number 

 of results is small. Where a larger number of results is available, more 

 satisfactory measures can be used. The most direct for the purpose of 

 showing the dispersion of soil analyses appears to the writer to be the 

 interquartile range^. The method of obtaining it may be illustrated 

 from an actual example. In the case of the Anglesey medium loam 

 the percentages for fine sand in the soil are arranged in ascending order 

 of magnitude. There are 26 observations. The lower quartile is given 

 by the seventh value from the bottom of the list and the upper quartile 



^ Hall and Russell, this Journal, vol. iv, pt. 2, p. 193; and Russell, Journal of the 

 Bd. of Agric. July, 1916. 



^ A large number of the mechanical analyses were performed by my laboratory 

 assistant, Mr H. H. Hughes, whose loyal help is here acknowledged. 



* G. Udny Yule, Introduction to the theory of Statistics, 1912, p. 147. 



