42 Mr. D. Thoday. Ex2:>erimental Researches on [June 11, 



It would be easy to devise improvements of the templet method which 

 would ensure very accurate cutting. By using thick templets with vertical 

 edges and a cutting instrument with flat blade fixed vertically, capable of 

 being brought down simultaneously on to the whole length of the required 

 cut, all the. errors that have been mentioned would be practically abolished. 

 By such an instrument, too, a number of leaves could be manipulated at the 

 same time without any sacrifice of accuracy. 



It may be said in conclusion, therefore, that the errors of the templet 

 method, though appreciable when the method is used in its original form, are 

 capable of almost indefinite reduction. 



The Rotating Punch. 

 A modification of the templet method has already been devised for the 

 purpose of cutting small pieces from between veins, while at the same time 

 avoiding the objections to using small templets by ensuring extreme accuracy 

 in cutting. For the details of the construction of the instrument. Dr. F. F. 

 Blackman was responsible. It consists of an adapted watchmaker's drill 

 fitted with a circular ciitter revolving on its own axis, and accurately turned 

 in situ for this purpose. The cutting tulje slides freely in a vertical direction, 

 and the method of procedure is to set it in motion by means of a water 

 turbine and then bring it down on the leaf, so cutting out a disc equal in 

 area to that enclosed by the cutting edge. This can be repeated rapidly, a 

 number of discs accumulating in the interior of the tulje. A lateral slit 

 facilitates their subsequent removal. 



T]ic Planimeter Method. 

 The planimeter method was that adopted Ijy Brown and Escombe as a 

 general metliod. 



They estimate the error in llieir iiie;isuieinent of leaf-piiiits hy a planinietor as "well 

 under O'l per cent."* This is ai)parently to be taken as holding for areas as small as 

 40 sq. cm.,t although usually tlie areas measured were of the order of 100 eq. cm. 

 The absolute error corresponding to their estimate must therefore be well under 

 0"05 sq. cm. 



Where this method was adopted in the present investigation, the instrument used 

 read by a vernier to O'l sq. cm. I'nder favourable circumstances, the maximum error 

 found, due to inaccuracy in following the bounding line, was ± 0"1 8(}. cm. in measuring 

 an area of 21 b s(). cm.\ I am therefoie inclined to regard 1'rown and Eacombe's estimate 

 aa too low. 



* Loc. cit., p. 59. 



t Cf. loc. ( it., ji. «0, Table IX. 



I Besides this error otliers were found : on testing the insti ument the graduation 

 proved inaccurate, and in addition different results were obtained according to the average 

 angle between the arms It is thus very necessary to test a ])laiiimeter carefully before 



