149 



usually in September, the occurrence of minimum levels bein^ 

 distributed among the months as follows: In the records at 

 Copperas Creek, eight occur in September, six in August, five 

 in October, three in July, and one each in November, December, 

 and June. The lowest level was recorded at Copperas Creek in 

 1879, when the low-water mark was established. It has not 

 again been reached, owing since 1889 to the dam at LaCrange. 



The low-water period is often one of marked stability as 

 contrasted with other parts of the year, the total movement of 

 river levels falling to 0.10 foot per month in November, 1893, 

 and frequently amounting to less than one foot in September, 

 October, and November, while even this movement is probably 

 caused to a considerable extent by the operation of the locks 

 and by changes in the direction or force of the wind. The stage 

 of extreme low water is followed by a gradual but very slight rise 

 during the fall months, which cannot be attributed entirely to 

 rainfall since, as shown in the table on page 126, these are 

 months of lessened precipitation. This increase is well shown in 

 the hydrographs of 1893 and 1897. It seems more probable that 

 with the falling temperatures the loss by evaporation, both from 

 the river and its tributaries, is sufficiently lessened to account 

 for this slight rise in levels, amounting in most cases to about 

 one foot. 



This low-water period is frequently interrupted by minor 

 fluctuations, some of which appear at or subsequent to the 

 autumnal equinox. These fluctuations are due to heavy sum- 

 mer rains, and usually appear suddenly and decline with almost 

 equal abruptness. They rarely rise to eight feet and are usually 

 below five; they thus do not cause overflows, and affect only 

 those bayous and lakes which maintain connections with the 

 river at low-water levels. Their duration is short also, being 

 but a week or ten days, rarely a fortnight. In 1896 there was 

 a repetition of such rises of more than usual prominence and 

 duration, giving a unique character to the liydrograph of that 

 year. The equinoctial period, marked by the slightly increased 

 rainfall of September, is not ma^rked in the average hydrograph 



