METHODS OF INVESTIGATION 183 



crept into our beliefs about cotton may be traced to this 

 handling of lint only, though, if it were practicable to 

 adopt the method of measuring single fibres initiated by 

 O'Neill, many of these could be eliminated. 



The three methods are as follows : 



(a) Single-Fibre Measurements. Lint hairs whose ends 

 are visible on the outside of a loose heap of lint are pulled 

 out one by one, laid down with the wet finger on black paper, 

 and measured with dividing compasses. (In all measure- 

 ment work it is necessary to use dividers which are after- 

 wards placed on a scale, since subjective error comes in if 

 the scale is directly applied.) Provided that microscopic 

 examination is included in the programme to insure that 

 each fibre is unbroken, the result in good samples has 

 a probable error for single fibres of about 10 per cent. 

 O'Neill's original figures for Sea Island work out at about 

 8J per cent. Thus the measurement of 100 fibres gives 

 a probable error of 1 per cent., or 25 fibres 2 per cent. We 

 shall see below that the same precision may be obtained 

 by combing six seeds only as by measuring 25 fibres. 

 There can be no doubt as to the relative ease of manipu- 

 lation; single fibres are quite easy to handle in a good 

 light, but any prolonged work with them strains the 

 eyes severely, as the author knows only too well. 



(6) Pulling. The grader takes opposite sides of a lump of 

 lint in the whole grip of each hand and draws them apart ; 

 he then grips the projecting fringe of one half between 

 finger and thumb along a straight line and draws again; 

 the distance from the grip to the end of the fringe gives 

 the length, and the "hardness of the edge" gives the 



