1905.] 



On the Retention of Bases by the Soil. 



13 



others, the rate of loss is then based upon the difference between the earliest 

 and latest samples from each plot. 



B. Loss of Calcium Carbonate on Unmanured Land. 



It will be convenient to consider the unmanured plots together, so as to get 

 an idea of the initial loss of calcium carbonate to the soil when there are 

 no disturbing influences introduced by the manure. All the unmanured soils 

 show the same general characteristics — a comparatively large amount of 

 calcium carbonate in the top 9 inches, varying from nearly 6 per cent, on 

 Agdell Field down to little more than 2 per cent, on Hoos Field. The second 

 depth, however, contains as a rule between 01 and - 2 per cent., only in a 

 few special cases does it rise above the latter figure ; in the third depth the 

 proportion is 0*1 per cent, or less. As the Eothamsted fields are situated on 

 almost the highest levels of the chalk plateau, it is difficult to suggest any 

 reason for the restriction of the calcium carbonate to the surface soil other 

 than its artificial origin, and this conclusion is confirmed both by the irregu- 

 larity in the amount found in different fields and by its absence where the 

 land has not been under cultivation. Samples were taken from three places 

 on the adjoining Harpenden Common, where the same class of land occurs 

 at approximately the same level as the experimental fields, but which has 

 never been in cultivation and now carries a growth of poor grass, gorse, and 

 bracken. 



The results : — 



Soil of Harpenden Common, 1 to 9 inches deep... - 210 p. c. CaCOa. 



10 to 18 „ ... 0136 



show that initially the soil of the Eothamsted Estate must have been practi- 

 cally devoid of calcium carbonate. 



The successive determinations of the amounts present in the soil of the 

 cultivated fields give the following rates of loss per acre per annum for 

 the unmanured land: Broadbalk 800 lbs.; Hoos, 1000 lbs.; Agdell, 922 

 and 938 lbs. ; Little Hoos, 1046 lbs., or an average loss of over 944 lbs. 

 per acre per annum due to percolating water only. These rates are all 

 probably too high, because of the gradual exhaustion of the organic matter 

 and the resultant consolidation of the soil which has affected the later 

 samples. But even after making all corrections that are possible for this 

 source of error, the rate of loss would still amount to 800 lbs. per acre. 



These estimates may be to some extent checked by certain determinations 

 made in 1896 — 8 of the amount of calcium compounds present in the water 

 percolating through the drain gauges at Eothamsted. These drain gauges are 



