JANUARY 4, 1884.] 
sow, Ziwolka, and Moisseew (1882-39), and a few 
notes by Hofer and Nordenski6ld, and from his own 
observations, Grinewetzky gives the following sketch 
of the south ,island of Novaia Zemlia. It may be 
divided into three parts. The northern lies between 
Matotschkin Shar on the north and the Pukowaja 
River on the south: this part has the highest moun- 
tains (four thousand feet), forming isolated groups 
rather than ranges. The central part, extending to 
the Karelka and Belushia, Rivers, has five or six 
parallel ridges, rnnning generally north and south; 
black slate is common; and the watershed is about 
seventeen miles from the west coast. The southern 
part is a rather low plateau: the Goose Land (Gusi- 
waya Zemlia) isincluded in this part, which is free from 
snow by the end of June, andin July has arather rich 
vegetation, especially on the gently sloping ground. 
Dr. Grinewetzky also expressed the opinion that 
the wild reindeer of the northern island belong to a 
totally distinct sub-species from those of the south- 
ern island. 
LOSS OF NITROGEN FROM ARABLE 
SOILS. 
THE renewed attention of agriculturists has of late 
-been drawn to the question of the nitrogen supply 
of cultivated soils. On the one hand, Schulz, in 
Germany, claims to have brought about a gain of 
nitrogen on a sandy soil by means of the cultivation 
of Jupines, and manuring with kainit. On the other 
hand, Lawes, Gilbert, and Warington,! in England, 
have published results which show that a very 
considerable annual loss of nitrogen occurs in the 
drain-water of cultivated fields; and experiments by 
Dehérain,? in France, show, according to his inter- 
pretation of them, an alarming decrease in the total 
nitrogen of the soil in the course of a few years, and 
in spite of abundant manuring. 
Schulz’s experiments have added nothing to our 
knowledge of the natural supply of nitrogen to the 
soil, and it is not proposed to consider that topic 
here. The results of Lawes, Gilbert, and Waring- 
ton, and of Dehérain, however, have attracted much 
attention. If they are to be accepted without re- 
serve, they lead to the conclusion that the fertility of 
our cultivated fields, so far as it depends upon their 
nitrogen, is being removed in the drainage-water, or 
in other ways, at a comparatively rapid rate. 
The instigation to Lawes, Gilbert, and Waring- 
ton’s experiments was given by the observation, that, 
in the field-experiments carried on for a series of years 
at Rothamsted, scarcely a third of the nitrogen of the 
manure was found in the crop under the most favor- 
able conditions, while, in those cases in which no 
mineral manures were applied, the deficit was much 
greater. The most obvious conclusion was, that there 
must be a great loss of nitrogen in the drainage; and 
experiments were instituted to test this idea, Their 
earlier experiments were with three lysimeters. Ex- 
cavations were made under and around an area of 
1 Journ. roy. agric. 30c., xvii. and xviii. 
2 Annales agronomique, Viii. 321. 
SCIENCE. 17 
a thousandth of an acre. The mass of soil thus 
isolated was supported by perforated iron plates, 
and surrounded by masonry, thus leaving the soil 
with its natural structure. The quantity of water 
percolating through this soil has been determined 
since 1870; and since May, 1877, its content of 
nitrates has been also determined. The soil was 
uncultivated and free from vegetation. Numerous 
interesting facts are disclosed by these determina- 
tions, but that which now interests us chiefly is the 
quantity of nitrogen found in the drain-water. This 
amounted, in the average of four years, to 46, 36, and 
44 pounds per acre, at depths respectively of 20, 40, 
and 60 inches. 
Subsequently the same experimenters have de- 
termined the nitrates in the drainage-waters from 
their experimental wheat-field, each plot of which is 
drained by a single lateral at a depth of 24 to 30 
inches. Having no means of measuring the drainage, 
the authors take,:as the basis of their calculation of 
the loss of nitrogen, the amount of drainage-water 
yielded by the 60-inch deep lysimeter at the same 
time. On this assumption, the annual loss of nitrogen 
varied from 15 and 16 pounds per acre, on unmanured 
plots, to as high as 74 pounds per acre. 
It is greatly to be regretted that the authors were 
not able to measure the drain-water in these experi- 
ments; for the method which was adopted to supply 
the deficiency leaves much to be desired. The soil 
in the lysimeter was uncultivated and bare of vege- - 
tation: that of the wheat-field was cultivated, and 
bore crops of wheat varying considerably in amount. 
Both these circumstances affect the amount of drain- 
age-water. Cultivation, especially of a clay soil such 
as that-at Rothamsted, may affect very markedly the 
ease with which water passes downward through it, 
the amount of water which it can retain in its inter- 
stices, and the rapidity of evaporation from its sur- 
face. The growth of vegetation exerts a still greater 
effect on the movements of water in the soil. It has 
been shown by numerous observers, that much more 
water evaporates from a soil covered with vegetation 
than from a bare soil, and that cousequently much 
less of the rainfall percolates through the soil. The 
diminution of the drainage-water in this way has 
also been directly proved by Wollny. Furthermore, 
the various plots in these experiments carried un- 
equal quantities of vegetation, so that the amount 
of evaporation due to this source must have been 
unequal also. It appears, then, in the highest de- 
gree improbable, that the quantity of drainage-water 
actually was the same for each plot as was assumed, 
and unlikely that it was as great as was assumed. 
When we add to these considerations the fact, that 
it is uncertain whether the soil of the lysimeter 
represented an average of the soil of the field, and, 
further, that all errors of the lysimeter are multiplied 
a thousandfold when the results are expressed per 
acre, we are forced to the conclusion that the figures 
given for the total amount of drain-water, and con- 
sequently those also for the total loss of nitrogen in 
this way, can be, at best, only approximations, and 
are most likely too large. 
