APPENDIX IV. 



THE ACCUEACY OF THE OBSERYATIOXS. 



(1) Calculation of the error. If we assume that no constant 

 error is involved in the use of the quantitative net, and that the 

 plankton is distributed with absolute uniformity, then the experimental 

 error of the observations may be calculated by statistical methods. A 

 number of such determinations have been made and are summarised 

 by Schiitt (in Analytische Plankton-Stiidien, Kiel and Leipzig, 1892). 

 The most obvious manner of estimating the error is by making parallel 

 hauls with two nets couj^led together on the same line and lowered and 

 hauled at the same time. Schiitt's Table 7 gives the results of six such 

 experiments made during the Plankton Expedition. These are : 



The average error is thus 5-4 7o- Calculated by the method of 

 least squares it is 6-8 '^/o- The probable error is however 4-5 o/^. 



The results of 54 test hauls made by Hensen are also quoted by 

 Schiitt in Table 6. In these trials the same net was hauled twice in 

 succession in the same place, and as nearly as possible from the same 



