— 95 — temperature accompanied by extreme aridity, in September and October as has sometimes been recorded. Such weather is usually coupled with 'a marked shedding of leaves and bolls as a natural response to the demand for a reduction in the transpiration surface, and accounts for a great discrepancy between flower-formation and boll-development and a consequent serious reduction in the yield of the crop.

The problem is, therefore, whether it be preferable to plant early and expose the tender seedlings to hot, dry winds and have the plants well grown by the time the cold season arrives, or to wait until the " habub " season is over and risk later development being checked by abnormally cold winds. Until this point is settled an element of speculation must always attend the planting of cotton, as obviously it is impossible to forecast the weather for months ahead. In this connection it should be noted that plantations in the Khartoum Province are undoubtedly more favourably situated, and sufEer less from excessive changes in temperature and humidity, than estates in the Berber area where remarkable fluctuations in crop returns clearly reflect the importance of the weather factor.

The 1913-14 season on the farm was marked by a premature cold spell in winter, and the early-planted cotton consequently had the advantage. The 1914r-15 season, however, was characterised by its high humidity and mild winter. Later-planted cotton which had escaped the " habubs " was therefore greatly superior.

It is unfortunate that all our experiments during the 1913-14 season were rendered unreliable by the breakdown of the pumping plant and the consequent shortage of water. Further, that the Times of Planting Experiment of the 1912-13 season was not commenced until July 10th, thus giving us no check on cotton planted last season on June 1st, 10th, 20th and 30th.

The experiment is carried out annually over an area of 10 feddans, which are sown consecutively at intervals of ten days.

From data at our disposal the best average crop is obtained from cotton planted on July 10th (vide Chart 3). It is further obvious that cotton planting after August 20th is out of the question. It is to be remarked that the feddans which produced the best yields last season were sown just as the bulk of the rain was falling.

There were rains on July 28th and 30th and Augast 1st, 2nd, 5th, 8th, 11th, 21st, 23rd, while the cotton was planted on July 20th and 30th and August 10th and 20th. The crop received no checks, as there were neither early cold winds nor was the young growth subjected to the withering influences of high temperatures, extreme aridity, or " habubs."

Such conditions, of course, represent the ideal, and reason seems to indicate that it is safest to sow cotton during June and the first half of July, as the risks attendant on later-sown cotton more than counterbalance the possible benefits to be derived from deferred planting.