38 DISCOVERY REPORTS 



Oithona frigida: — A line: 901 3064 2031 1339 

 C line: 1491 1844 2045 3368 



SG/FI line: 3753 7211 1853 3524 793 

 Whilst the laboratory error is of the order 15-25 per cent, considering the errors in 

 the field, demonstrated by the results at St. 41 above referred to, we are not justified 

 in drawing any other conclusion than that they are fairly evenly distributed and occurring 

 in numbers of two or three thousand per 50 m. haul. 



Drepanopus pectinatus: — Aline: 124 2563 8 65 



C line: 2639 6 7 3 



SG/FI line: 2 0000 



Here, where again the laboratory error does- not exceed 25 per cent, in spite of 

 possible errors in the field, we can at once discern big differences between different 

 points in the area, differences which, as we shall see, have a definite meaning. 

 Calanus propinquus : — A line: 29 1 5 14 



C line: 16 34 55 139 



SG/FI line: 39 25 820 gn 1360 



Here whilst the laboratory error did not usually exceed 30 per cent, but might reach 

 50 per cent, the errors in the field will probably not allow us to distinguish between the 

 majority of stations, but we can fairly record a marked difference between those in 

 italics and the rest. It is this nature of difference that we are concerned with recording. 



It may be asked why do we not express the estimated numbers in round figures if we 

 cannot distinguish between 820 and 911 or between 39 and 25. For two reasons: 

 firstly, any rounding of the figures is going to introduce a further possible and un- 

 necessary error, and secondly, whilst the error in the field due to differences in time 

 counts against the exact comparison of different stations, it does not so count against 

 comparison of hauls at different levels at the same station where these hauls are taken 

 immediately following one another. Here only the laboratory error is involved, and 

 whilst we are again only interested in the big differences it is well not to increase un- 

 necessarily the laboratory error by arbitrary rounding. As an example of such depth 

 distribution let us take the SG/FI line for Calanus propinquus in greater detail, showing 

 the number per 50 m. haul for the top 250 m., as sampled by the routine N 70 V nets 

 taken at 250-100 m., 100-50 m. and 50-0 m. The depth distribution at the five stations 

 would appear as follows: 



50-0 m. 34 44 3526 1919 6530 

 100-50 ,, 16 50 548 951 6 



250-100 59 7 9 5 61 88 



Here, in spite of the laboratory error, we clearly see the nature of the important 

 differences between different levels ; but it is well not to increase that error further by 

 arbitrary rounding. Again, if these reasons had not been good ones, having indicated 

 the values to be placed upon the figures, we should have considered time too valuable 

 to be spent in rounding them all simply to save us from the possible criticism that they 

 appear ridiculous. 



