277 



the average, in all probability, fall within ± 10 per cent. 



Our chronological series of collections affords a few instan- 

 ces of catches under somewhat stable conditions of river levels 

 and temperature, and at intervals so short that they may be 

 utilized as tests of local distribution within certain larger limits 

 of error, since the utilization of such data introduces the er- 

 rors resulting from changes of chemical conditions due to rot- 

 ting of sewage, and from growth, reproduction, and destruction 

 of the plankton in the interim between collections. The follow- 

 ing tabulated instances (p. 278) from Table III. and Plates X. 

 and XI. may be cited as throwing light on this question of local 

 distribution along the length of the stream. 



The fourteen groups of collections were selected with refer- 

 ence to stability of conditions, therefore in falling or low water 

 and in periods of relatively even temperatures. Inspection 

 of the tables and plates above referred to will show that the 

 selection has not been made so as to eliminate wide varia- 

 tions, and it may therefore be regarded as fairly typical. The 

 periods included, range from 2 to 15 days in extent, and 

 upon estimated rates of current the several tests include plank- 

 tons taken at intervals in reaches of channel water from 24 to 

 252 miles in length. The average departures from the mean, 

 range from ±0 to ±29.8, and yield a grand average of ±14.1. In 

 view of greater number of catches averaged and extended 

 time element involved, these results compare very favorably 

 with those derived from Reighard's data and Apstein's results. 

 The probable error resulting from variations in the longitu- 

 dinal distribution under stable conditions seems to be less than 

 ±15 per cent. 



An inspection of Table III. and Plates X.-XIII. will show 

 that in the case of invading flood waters the departures from 

 the mean of catches at similar intervals would be considerably 

 greater than the averages above computed. Also, that in case 

 of plankton pulses in stable conditions — for example in Sep- 

 tember and October, 1897 — collections at weekly intervals may 

 exhibit departures in excess of ±50 per cent. It is evident, how- 



