190 THE DEVELOPMENT OF RAW COTTON 



in keeping numerical records of some sort; and since it 

 was obviously impossible to set up spinning tests in 

 Egypt, data as to breaking strain of single fibres were 

 better than nothing. 



From Mr. Hughes' data it appeared that with proper 

 sampling the probable error of single fibres was 15 per 

 cent., so that a test made on fifty-six fibres would have a 

 probable error of 2 per cent., or, in other words, that 

 strength-testing could be made as accurate as length- 

 testing without using an enormous number of fibres. 

 Even at five minutes per fibre this meant an hour 

 for each sample, and, as the author had one series re- 

 quiring testing which alone consisted of sixty samples, 

 he cast about for some method of speeding up the process. 



Automatic Tester. The outcome of some weeks of 

 instrument-making in spare moments was a home-made 

 machine which tested fibres one by one automatically, 

 at the rate of one in twenty-five seconds, and single-fibre 

 testing became practicable on a large scale. At the same 

 time the instrument is or, rather, was a purely labora- 

 tory appliance, since it necessitated the native lab-boy's as- 

 sistance to mount up the fibres on cards somewhat similar 

 to those used by Mr. Hughes ; these cards were loaded in a 

 magazine, and the author's part in the testing consisted 

 in aligning the magazine to the tester proper, cutting 

 the cards and pulling over a switch. The magazine then 

 swung in to the. testing-points and placed a pair of half- 

 cards upon them, with 10 mm. of fibre connecting them, 

 moved away, and stopped. The tester then strained the 

 fibre at a constant rate against a spring- balance which 



