A FORTNIGHTLY REVIEV/ 



OF THE 



LIBRARY 



NEW YO' 



IMPERIAL DEPARTMENT OF AGRICULTURE FOR THE WEST INDIES. TaZI 



Vol. VIII. No. 198. 



BARBADOS, XOVEMBEli 27, 1909. 



PmcE Id. 



CONTENTS. 



Page. 



Bo.ik Siielf 383 



Brii- .11 Com Market in the 



Tnitod States ."^81 



Cocoa-nut Palms, Care of 373 



Colonial Fruit Show in 



December 382 



Cotton, Experimental 



Breeding in India ... .37() 



Cotton Notes : — 



Selection of Cottm 

 Varieties for Unifor- 



111 ily 374 



West Indian Cotton ... 374 



Department News 375 



Departmental Reports : — 

 Miiiitserrat, Botanic Sta- 

 tion, etc 382 



Biitish Himduras, Bot- 

 anic Station 382 



Fibres, Hiind- stripped and 



Machine-stripped ... .376 



Field Experiments, Value 



of Results ... 3G9 



Fruit Trade, Sicilian, in 



1908 377 



Page. 



Fungus Notes: — 



How Fungicides should 

 1)3 Employed 



Gleanings 



Grass ( )ils in Indo-China 



Ground Nut, The 



Imperial Blalaria Confer- 

 ence 



37! I 

 38(( 

 377 



Insect Notes : — 



Natural Histi >ry of 

 Insects, Part III ... 



Market Reports 



Notes and Comments ... 

 Perini Fibre I'lant, The 



Rice in British Guiana ... 

 Rubber Tree, The Tonkin 

 Rubber Trees, Mamirial 



Experiments with ... 

 Selection and Breeding, 



Aid in 



Students' Corner 



Sugar Industry : — 



Sugar Factoiy Results in 



Java 



37 



378 

 384 

 370 

 375 



383 



377 



37(i 



381 



371 



The Value of the Results of Field 

 Experiments, 



is a well recognized fact thiit, in e.xperi- 

 \f&>' ments which have to be math; in the field 

 ^•^■^=&l there is a large number of sources of error 

 which arise directly from the conduct of such experi- 

 ments on a scale which precludes the possibility of 

 complete control. The duty, therefore, of the investigator 

 is to devise ways by which such errors may be minimized, 



as fir as possible, and by which he ma\- ascertain their 

 magnitude, and so put himself vj a position to make 

 allowance for them whei; deducing results. An able 

 paper dealing wit'i this subject, by A. D. Hall, M.A., 

 F.R.S., Director of the Rothamsted Experiment il 

 Station, ;ip|:ears in the Journal of the Board of 

 Agricaiture, August 1909, in which the following are 

 the main points that are brought forward. 



No one who does work which entails the measure- 

 ment of ijuantities, whether dimensional or in relation 

 to weight, e.xpects to obtain absolute accuracy. A joiner 

 or carpenter is satisfied if his measurements are wiihin 

 an eighth or a si.Kteenth of an inch; the mechanical 

 engineer reijuires accuracy within one-thousandth of 

 an inch, wliile the scientist, who makes observations 

 with the ud of a microscope, must possess certainty 

 within the simpler fractions of one twenty-five 

 thousandth of an inch. For all these, there is a method 

 available for increasing the accuracy of their work: that 

 is, not to rely on the result of one observation, but to 

 repeat it several times, and to take the average as being 

 sufficiently correct for theii purpose. They can do 

 more than this. By a suitable mathematical process, 

 they can determine, from the direct measurements, 

 what the error in the final result is likely to be, and 

 will thus have the means of allowing for it in 

 subsequent work. 



An illustration of this may be useful, for the better 

 apprehension of the principle. If, for example, the area 

 of a piece of land is found, by several measurements, to 

 be near 1840, 1S4'3, 1S3-.5, 184-G and 183-3 square 

 yards, the average, 184 square yards, which is easily 

 deducible in the ordinary way, may be taken as the true 

 result. Further, the ' probable error ' of this result, that 

 is the limiting quantity by which it is probably incor- 



