The harvest of the cotton crop usually begins in extreme southern Texas about July 1. By the middle of August picking is in progress throughout the southern portion of the Gulf Coast States and during the first decade of September this work extends to the more northern districts. Owing to the slow and tedious process of picking, however, harvest is extended over a period of several months, even after the plants are all fully matured, and is not usually finished until well into the winter season.
In conformity with an act of Congress, the Bureau of the Census, Department of Commerce, issues during each harvest season 10 preliminary reports of the amount of cotton ginned to specified dates at approximately semimonthly intervals. They are based on data collected by local agents of the Bureau who canvass the ginners. These reports are considered of great value by those interested in cotton production, as they not only place in possession of all concerned reliable information as to the rapidity with which the crop is being harvested, but by reason of affording deductions as to the amount of the final output made possible by a careful comparison of current reports with those of previous years. The earliest official estimate of the amount of cotton produced is made by the Bureau of Crop Estimates, Department of Agriculture, about December 12, of each year.
While the amount of cotton ginned to a specified date or during a given ginning period has some value as a basis for forecasting the final output, the data are often misleading when the weather factor, which so largely influences the progress of harvest, is ignored.
The rapidity of harvest varies greatly from year to year, depending principally on the earliness or lateness of the crop and the weather conditions prevailing during the harvest season. The relative amounts ginned during the earlier ginning periods in different seasons depends principally on the earliness or lateness of maturity, but later in the season the rapidity of harvest is determined by the prevailing weather conditions.
The relation between the amount of cotton ginned during November and the prevailing weather of that month has been mathematically determined as a basis for forecasting the final output.
By November 1 the cotton crop has practically matured but, on the average, 37 per cent of it remains to be ginned on that date. Fifty-seven per cent of that remaining unginned on November 1 is ginned during the month of November, on the average. These latter percentages vary greatly from year to year, depending on whether November happens to be favorable or unfavorable for picking and ginning. These variations have a very close relation to the number of rainy and cloudy days during the month, which affords a basis for computing the percentages for future years when the relation is mathematically determined. This has been done for the entire cotton belt and the results are given in the accompanying tables.
The closeness of relation between the average number of rainy and cloudy days in the cotton belt and the percentage of the cotton remaining unginned on November 1 that was ginned during November, is shown by the correlation coefficient of −0.91 ±0.03 (Table 6). This is among the highest coefficients on record where meteorological data are involved.
Ginning data for 15 years, 1905–1919, are available1 and computations have been made for this period on the basis as outlined, with excellent results. The final computations are shown in Table 8, from which it will be noted that the average error in the computed totals for the 15-year period was only 1.5 per cent, with an error as great as 2 per cent in only 5 of the 15 years, while it was less than one-half of 1 per cent in one-third of the years. By the application of the constants of the equation, shown in Table 7, to the November weather data in future years a reliable computation of the cotton crop can be made in less than 5 minutes after the amount of cotton ginned-to December 1 is reported by the Bureau of the Census. The final report of yield is not made by that Bureau until the latter part of March, or later.
Final data for 1920 had not been published when this article was completed.