A FORTNIGHTLY REVIEW 



OF THE 



IMPERIAL DEPARTMENT OF AGRICULTURE FOR THE WEST INDIES. 



r 



E 



Vol. XL No. 26-3 



BARBADOS, MAY 25, 1912. 



Pbice Id. 



CONTENTS. 



Page. 



A^'iicullimil C'dnfi'ii'iui- 



Piclurt- 



Aitiuultiiial Ci>-()j)er;itivf 

 Credit Snek'ties in 

 England and W:df>... 

 Cotton Notes : — 



Ci 't ton Indust vy in .Fa] lan 

 We.st Indian Cotton ... 

 Dei astmental Rf[inrts ... 

 Di'iiiinica. A^nciill iir.d 



riiin;ii.s> in 



Field Exi)eiinients, Intei- 

 pretation of Ue9nlt^ 



of, II 



Fungus Notes : — 



Inphioitel Canker 



K..>e Mililew 



Gleanings 



Goat lrii|iriivemeul A»oei- 

 ati<ms in Bel^imn ... 

 Index and Title P.ige 

 Insect Kotes : — 



A Disease o{ firassh(-(i 



The Banana Weevil 



Boier 



The C(.tt..n B..11 Weevil 



16S 



171 



166 

 166 

 167 



173 



145 

 161 



174 

 174 



172 



175 



168 



170 



170 

 170 



Page. 



Lima Bwm, Se'ef.idn (••r Kit 



Market Reports 176 



Nitiiigen-tixinsj' »rt;anisnis. 



Stimulation by Humus 169 



Notes and Comments ... 168 



Par.i nulil)er Seeds, Vital- 

 ity .<{ 16,T 



Ploughs. Method for Fa- 

 cilitating Removal of IC.'.i 



Poinsettia. Propagation of 167 



K.it Kxterminators, Value 



of 168 



Rnlilier, Meelianical Kx- 



traction of 16(; 



Soil Slerilizatidn. Partial, 



Use for 166 



.Sov Bean in England. 



Trials of ../ 160 



Students" Corner 173 



Sugar Lidu.stry: — 



Sugar-Beet in England 1(13 

 Tl-.e .\ntigua Sugar 



Faetoiy ... ' 163 



Trinid.id. Heeent Entomo- 



h.gieal Work in 169 



West Indian Products ... 175 



The Interpretation of the Results of 

 Field Experiments. 



IL 



5~5>^N the last number of the Agricaltaral Nmcs 

 consideration wa.s given to the ways in which 

 the results of field experiments may vary, and 



the need was shown of the possession of means for 

 determining the amount of variation that is likely to 

 takei)lace, under giTen circumstances, and for applying 

 this in the interpretation of results. The purpose of 

 the present article is to develop the subject, by setting 

 forth in a general way the measures that are av.-iilable 

 for the purpose under discussion. 



When an average is struck, of a large number of 

 results, it is generally obvious that, while most of the 

 results are near the average, a few of them, for reasons 

 that have been explained already, differ largely from 

 this. It i.s therefore expedient to possess a simple 

 means of ascertaining the amount of dependence that 

 may be placed on the average. Such a mean.s is 

 afforded by constructing what is known as a Freijuenc}' 

 Curve, in the following way. First of all, the results 

 are divided into classes, each class corresponding to 

 the same proportion of the terms in which the results 

 are expressed: for instance, each class might be 

 taken to correspond to a difference of one per 

 cent. The number of results falling into each class is 

 then ascertained, and a frequency curve constructed 

 b}" dividing a horizontal line into equal parts corres- 

 ponding to the different classes, erecting perpendicul- 

 ars from the middle point of each equal part, marking 

 on each of these perpendiculars a distance proportional 

 to the number of results falling into the corresponding 

 class, and joining all the points thus obtained. If there 

 is formed a fairly regular curve whose highest point, or 

 peak, is above, or nearly above, the point on the 

 horizontal line corresponding to the average, depend- 

 ence may be placed upon the latter as the true average 

 of the results. 



