the data of i . j^ ■.> . .. i. ^s only necessary to transform tit© abscissa to 

 the sine of the angle, replot as in figtire 8, and fit a straight line to 

 the now nearly rectilinear distributioa of values o Transferring "&© 

 straight line to the coordinates of figure 6 produces the curve drauro 

 there^, which is seen to describe fairly well the general course of change 

 in logarithmic plankton volume with time of day,, The scatter to both 

 sides of the curve is considerable and pre-sumably reflects chaEc©' Taria™ 

 tions, locality differences j and the diserapaneies between the arbitrary 

 sine function and the true nature of the diurnal change o 



That for this particular cruise there is a discrepancj' be- 

 tween the mathematical function and the empirical sit;uation is suggested 

 by the excess of points above the curve in the neighborhood of 0400 

 hours and the deficiencies in the neighborhood of 1600 hours » It appears 

 that the rise in the catches begins somev/hat later in the afternoon 

 and then rises more steeply than the corresponding sine cu2°V6, and th^t 

 the decline in the morning departs similarly <. We have not tested the 

 significance of these departures j, but similar graphs for the data of 

 other cruises do not show any consistency in the time or direction of dis° 

 crepancieso It is probable that the sine curve describes the general 

 diurnal fluctuation as well as maybe expected of any simple mathematical 

 function o It accounts for an important part of the variability in the 

 plankton hauls, as may be seen from the coefficient of determination 

 (r^) (Ezekiel 1950^ po 138)^ as given in table " for each of the several 

 cruises o 



Accepting, thenj, the sine function as describing the locus 

 of the diurnal cycle and the fitted regression line in the transfornsed 

 data as the quantitative effect of this cycle on the particular group 

 of hauls for cruise 8, the adjvistment to reraoTO the diurnal effect has 

 been computed as set forth in table 48 



Cruise 8 of the Manx ing was not designed j, however, to iB= 

 v6stigat.e the diurnal cycle, but was intended to dsterndne whether or 

 not there was any indication of a definite decline or irorease in the 

 plankton during the late summer spasoHo ■. sries of stations, rather 

 than a single one, v/as selected to give a hiore general saTrpling of the 

 plankton population and therefore general significs/nee to anjr change 

 which might be observedo An analysis of variance of the unadjusted 

 data (table 5)., with tv*c=>way classification (following Snedecor 1946), 

 indicates no significant differences among stations nor among weeks o 

 The mean square for discrepance is large, however, and tends to render 

 less sensitive the test of significance for the other mean square 

 values o Analysis of the adjusted data (table 5) shows a grfsatly reduced 

 total variance from that, of the unadjus^ced data, indicating that the 

 chief scarce of variation wa Ims of hauling,, The mean square for 

 weeks now emerges as larger . lat of the discrepance and just under 

 the 0o05 level of sigpif icanceo Although we still conclude that t?:ere 

 were not significant differences among stations cr among weeks during 

 the period of sampling, we now gain the idea that the weekly differ- 

 ence is much more important than the locality differences o Variation in 

 zooplanktoa volume v^ith. sampling time and the smoothing effect of the 

 sine adjustment on the data are illustrated in figure 9o 



18 



