36 THE RELATION OF DESERT PLANTS TO 



was circular and had an area of 23.76 sq. cm. The surface was some- 

 what below the edge of the dish, but this distance was the same in all 

 cases. The consistency of the 40 per cent sample was about that required 

 for modeling clay, perhaps somewhat too moist for such use; the 30 per 

 cent could still have been used for modeling; the 20 per cent sample was 

 cohesive under great pressure, while the 10 per cent sample was hardly 

 cohesive at all. The dishes stood in the laboratory and were weighed 

 at frequent intervals, readings being simultaneously taken on an 

 evaporimeter standing beside them. In order to eliminate the effects 

 of variations in the humidity of the air and of such slight air currents as 

 might occur in the room, the rates of water loss have been calculated 

 in terms of the evaporimeter rate for the same period. These rates 

 are presented in the form of curves in figure 7 (p. 31), the actual 

 quantities given being the quotients of the rate per hour of evaporation 

 from the soil divided by the corresponding rate per hour of the evapori- 

 meter. These ratios are plotted, as in other cases, at the middle of the 

 time periods which they represent. 



An inspection of these curves shows a curious initial behavior of 

 the 10 per cent soil. Its rate at the start was exceedingly high, but it 

 fell to a position below the other soils within the first three hours. 

 This is probably due to the fact that in this soil there was not enough 

 water present to even approximately fill the spaces, so that the actual 

 evaporating surface was very large, extending down into the soil for 

 some distance. As the surface soil dried out the checking of water 

 loss by the dry soil above became apparent in the rapid fall of the rate, 

 which continued to fall more and more gradually as the air-dry layer 

 increased in thickness. At the end of the experiment 6.95 grams of 

 water had been lost, or 69.5 per cent of the whole amount originally 

 present in the sample. 



In the case of the 20 per cent sample no such excessively high rate 

 was observed at the start, there being apparently sufficient water pres- 

 ent to close the spaces which were filled with air in the 10 per cent 

 sample. From the behavior of this curve it appears that with this 

 water content the soil can transmit water at a rate not very markedly 

 below the evaporation rate which prevailed at the time, and hence the 

 air-dry surface layer was very slow in forming. However, it did 

 gradually form, and after 22 hours this curve is seen to fall more rapidly. 

 At the end of the experiment 12.62 grams of water had been lost, or 

 63 per cent of the amount originally present. 



The 30 and 40 per cent samples show little tendency to form air-dry 

 layers; their curves do not descend markedly, and the rate of water 

 loss at the end of the experiment is approximately as great as at the 



