VALIDITY OF ANALYSES 3 , 



migration, unless some of the organisms are actually resting for part of the time on, or 

 within a few feet of, the bottom ; but whether this be so or not it would be a condition 

 which might apply to any stations taken from the bottom to the surface throughout the 

 survey. 



Since we know that many of our stations are separated by considerable intervals of 

 time we can only take into account the very big differences in the number of organisms 

 when comparing one region of the survey with that of another. This being so it is surely 

 unwise to attempt a closer approach to accuracy in the laboratory than that afforded by 

 taking only one fraction from each sample for analysis when we know that the very nature 

 of the samples we are examining will not warrant it. Further, under these conditions 

 we did not feel that we could justify the additional time taken by trebling the analyses 

 of some 800 samples when the masses of material from later surveys were mounting up 

 and awaiting the attention of our limited staff. It is our task to compare the broad 

 differences between region and region and season and season. 



The error in the laboratory due to taking only a fraction of the sample and estimating 

 the total numbers present in the sample is less for the smaller organisms, which are 

 more easily sampled by the pipette, than for the larger organisms. We will consider the 

 error involved in estimating the numbers of seven organisms of different sizes : Fora- 

 minifera, Oithona, small Calanoid Copepoda (including such forms as Ctenocalanus and 

 Drepaiiopus), medium-sized Calanoid Copepoda (including such forms as Metridia and 

 Clausocalanas), Calanus spp., Rhincalanus gigas, and Chaetognatha. 



Two samples were sub-sampled, each twelve times, to estimate the numbers of 

 Oithona, small Calanoids and medium Calanoids in them. They were one N 70 H net 

 sample from St. 121, and one N 70 V net sample from the station above referred to as 

 giving us evidence of patchiness: 41 D 150-100 m. The former sample was also used for 

 determining the error involved in estimating the number of Foraminifera. 



Foraminifera 



Foraminifera showed an especial tendency to be unfairly sampled, for owing to their 

 relatively high specific gravity and compact shape, they sink rapidly like grains of sand 

 and are liable to escape the stempel pipette. The larger colonial Radiolaria were also 

 difficult to estimate, since they tend to stick together so that they are not so evenly 

 distributed throughout the sample as other forms and consequently give a larger error. 

 Whenever possible these were picked out and counted separately before the sample was 

 fractioned, but where they occurred in large numbers the stempel pipette was used. 



The numbers estimated from sub-sampling the St. 121 sample were as follows: 



5-2 



