303 



ble that the rise of the 17th is responsible for the suppression 

 of a rising pulse whose culmination had not yet been reached. 

 The slight recovery in the following week is indicative of the 

 upward tendency in production thus interrupted. That con- 

 tinued low water in this month may attend great plankton 

 production is seen in the records of 1897, when the monthly 

 average (see table on page 292) is 8.83 cm. 3 to 1.52 cm. 3 in 1895. 

 On the other hand, in 1896 and 1898 the disturbed conditions, 

 with higher water and more current, are accompanied by 

 much reduced production, averaging only .38 and .69 cm. 3 



The last week in September witnesses the first stages of 

 marked decline in temperature from the well-sustained summer 

 heat of 75°-85°. The decline reaches 68° at the end of the 

 month. This phenomenon combined with the last flood to ac- 

 celerate and complete the decline of the September pulse 

 which had already appeared prior to the last flood. 



The October pulse has a duration of 29 days, — from the 2d 

 to the 30th, — and a maximum amplitude of .76 cm. 3 on the 

 11th and 15th, following a rise in nitrates and attending in- 

 creased sewage contamination (PI. XLIIL). Its mean falls on 

 the 18th, 31 days after that of the preceding pulse. This is a 

 month of stable low water approaching minimum levels, the 

 total movement in the pulse period at Havana being only 1.03 

 ft. The temperature in this period falls from 61° to 45°, and 

 this taken in connection with the fall of 11.5° in the preceding 

 week brings to bear upon the plankton production of this 

 month the cumulative effect of a decline of 27.5° and the results 

 of the low temperature of 45°. The consequence is that the 

 summer planktonts are killed off or reduced in numbers, and 

 the winter planktonts have not as yet had time or temperature 

 to reach any considerable development. The plankton produc- 

 tion is therefore low ; so low, indeed, that its pulse-like char- 

 acter is largely a matter of conjecture in the volumetric data 

 (cf. statistical data on PI. LI.). Phenomena of like character 

 are to be detected at corresponding periods of autumnal decline 

 in temperature in September-October, 1896 ; in October-Novem- 



