12 



However, for scmie weeks after ammcynium salts had been supplied 

 as manure to the Broadbalk Field, there was an enormous increase in 

 the amount of ammonia absorbed from the air, and the lower dishes 

 during this pericid absorbed much more than the upper one. This 

 may be taken as evidence that there will be some loss of nitrogen 

 to the soil as free ammonia, whenever ammonium salts are supplied 

 as a fertiliser to soils containing calcium carbonate, 



W, E, BKliNCHLEY. ** The Weeds of Arable Land m relation 

 io the Soils on which they grow.'* Annals of Botany. 

 1911, 25,155, 



This paper contains a description of the distribution of weeds 

 of arable land upon the following formations in Hertfordshire and 

 South Bedfordshire : — Clay-with-flints, Alluvium, Chalk, Gault, 

 Lower Greensand and Oxford Clay, A distinct association was 

 found to exist between some of the species and the soils upon which 

 they grew, but the determining factor proved to be the texture of the 

 soil, except in the one case of the calcareous soils derived from the 

 chalk. For example, Bartsia Odontites and Mentha arvensis were 

 confined to clay, Chrysanthemum segetum^ Unmex acetosella and 

 Spergula arvensis were confined to s^'ind, and are probably charac- 

 teristic of an acid soil condition, while the Bladder Campion, Silene 

 CucubaluSf Geranium pusillum and a few others were confined to 

 Chalk, A certain number of weeds, like Shepherd's Purse, Chick- 

 weed, Horsetail, Coltsfoot, etc, appeared to occur indifferently 

 on all soils. It is impossible to say how far some of the associations 

 are valid until the work has been extended to a number of similar 

 soils in different localities, 



W, B, Mercek and A. 1), Haij.. ' The Experimental Error 

 of Field Trials.' jour. Agric. Sci. 1911. 4, 107. 



This pillar contains a discussion of the results obtained in 1910, 

 when an acre of mangolds apparently uniform was harvested in 200 

 equal plots, and an acre of wheat was similarly divided into 300 plots, 

 from each of which the grain and straw were harvested separately. 

 The results are set out and discussed by various statistical methods. 

 They show that the probable error attaching to a field trial consisting' 

 of two 'plots alone is in the neighbourhood of plus or minus 5 per 

 cent., and this error, though it diminishes with tlie size of the plots, is 

 not greatly reduced when the plots are made larger than g^ of an 

 acre. By repeating the plots receiving any particular treatment and 

 scattering them about the experimental area, a considerable reduction 

 (an be effected in the experimental error, though an increase of the 

 number of plots above 5 is not attended by any great further 

 reiluction of error. The authors consider that the most satisfactory 



