— 91 — other times the water drains into the river. Some difficulty has been experienced in preventing the formation of large cracks on the edges of the drain when the land surrounding it is not continually watered.

These drains have served their purpose admirably. The salt content of the soil is being steadily reduced and seepage water is at once carried off.

Rotation of Crops.—A simple three-course rotation covering a period of two years has been adopted both in the case of the comparative experiments and on the revenue area. Cotton occupies the land from July until April. It is followed by a summer crop of legumes, of which lubia afin (Dolichos lablab) gives the most satisfactory results in our heavy clay loams. The bean is fed, cut, or ploughed under in the autumn in preparation for the wheat crop, the planting season for which extends from middle November to middle December. With a growing period of some 120 days, the cereal is off the ground in ample time to permit of thorough preparation for the second planting of cotton in July.

Repeated inter-row cultivation of the cotton renders this an excellent cleaning crop, while the dense growth of the lubia still further checks weeddevelopment. The restorative character of the latter staple is dealt with exhaustively in connection with the problem of the nitrogen supply. The desirability of perennial irrigation and the limitation of bare fallows is recognised by this arrangement, and where circumstances permit a second leguminous catch crop is taken between the wheat harvest and cotton planting.

Scheme of Experiments.—The two-year rotation necessitates the division of the experimental area into duplicate series of hoshes, north and south, on each of which five distinct investigations are continuously conducted with the three staple crops, viz. :—I. Manuring. II. Distance of Planting or Quantity of Seed. III. Times of Planting. IV. Duty of Water. V. Methods of Cultivation. In the southern section an additional hosh has been devoted to a special investigation with green manures. Thus, in any season, cotton experiments are to be found in one area, and duplicate tests with wheat in the other. Manuring, Duty of Water and Cultivation Experiments are continuous on the same plots in order that cumulative effects may be recorded. The Spacing and Times of Planting plots, however, are reversed iij series each year to eliminate as far as possible the influence of soil irregularity. In the former cases, a measure of this factor is ascertained by the inclusion of numerous check plots.

The relative advantages of the chequer-board system of plots, as applied to certain field experiments, have not been lost sight of, but with a very limited supply of intelligent and trained labour, it has appeared that the risk of recurring mistakes would be greater than the margin of error obtaining from long and narrow parallel plots with an equivalent number of check, or untreated beds. The remarkable uniformity in chemical and physical characters of the soils under test, coupled with an independent system of laboratory control, whereby