1922] Allen: Studies on Diatoms and Dinoflagellates 371 



Such numerical differences obviously suggest the possibility that 

 diatoms generally tend to reproductive activity at night rather than 

 in the daytime when their energies are occupied in photosynthesis. 

 It also appears that dinoflagellates are not thus inclined. This differ- 

 ence in the groups strongly indicates that their characteristics are so 

 different that they cannot properly be considered together as a photo- 

 synthetic assemblage. It also indicates the justice of Kofoid and 

 Swezy's assertion (Kofoid and Swezy, 1921) that the holophytic 

 aspect of the activities of dinoflagellates has been over-rated. 



In addition to these points there are the repeated evidences that 

 distribution of organisms in sea water is not uniform. There were 

 very few days in the whole year in which catches twelve hours apart 

 showed approximately the same numbers and these few days were all 

 found to be in periods of very low production. 



The Twentt-foub-Houb Series 



In working over the records of catches at both La Jolla and Pt. 

 Hueneme it was found necessary to use some unit of time greater 

 than the day and less than the month. In some ways five-day units 

 seem to be most satisfactory, but they are not in common use and on 

 account of omission of Sunday collecting at Pt. Hueneme this unit 

 could not be applied. It was therefore decided to adopt the week as 

 the standard unit in this kind of discussion. This unit is_ also desir- 

 able because of its use by our hydrographic workers and by the 

 United States Weather Bureau. 



Seasonal Variations in Numbers 



One of the first questions to be raised concerning a series of catches 

 running over a period of a year is that about seasonal variations, the 

 more obvious of which may be expected to be variations in numbers 

 and variations in species or representative forms. As to the first it 

 is generally understood that maximum production occurs in the late 

 spring or early summer and that there is also a rise of production in 

 autumn after a decline in late summer. While our 1920 evidence does 

 not contradict this view, it does modify it to a considerable extent, in 

 that the numerical distribution is not thus limited at either station. 



