536 PROFESSOR KNOTT ON THE STRAINS PRODUCED IN 



above the axis in the vicinity of Field 300. In this instance there are consequently 

 three changes of sign: (1) from positive to negative in Field 93, (2) from negative to 

 positive in Field 290, and (3) from positive to negative in Field 370. 



This tendency for the " iron-cap " curves to lie a little lower than the " brass-cap " 

 curves may admit of several explanations. It may, in the first place, be accidental ; for 

 the difference between the "brass-cap" and "iron-cap" curves for one tube is no 

 greater than may occasionally be obtained between two "iron-cap" curves obtained 

 from different experiments. It was, indeed, practically impossible to obtain exactly the 

 same series of readings from a second experiment with the same tube set up afresh in 

 (so far as could be judged) exactly the same state. The final screwing in of the cap 

 seemed in some cases to have a slight effect — the harder the screwing in the smaller the 

 volume change. Now the iron and brass caps will differ in the grip they take of the 

 iron tube, and it is quite conceivable that this may give rise to a difference in the 

 volume change under magnetisation. 



Then, again, there may be some lingering trace of the false effect discussed in section 

 3, so that the iron cap, though screwed in as hard as possible, may nevertheless experience 

 a slight bodily displacement towards the tube. This would produce a slightly increased 

 diminution or a slightly diminished increase of volume. 



And, finally, there is the question of the average magnetic magnetisation throughout 

 the tube, which will of course be greater and more uniform with the iron cap than with 

 the brass cap. The effect of the iron cap is, in fact, to lengthen the tube a little in one 

 direction. Consequently, the leakage of the lines of induction out of the tube will be 

 somewhat less when the iron cap is used.* In our ignorance of the true character of 

 the strain, which is probably very complicated, it is impossible to surmise in what 

 manner and to what extent a slightly more uniform distribution of magnetisation along 

 the tube will affect the volume change. As will be seen below, the effect of substituting 

 the brass cap for a cap made of the metal of the tube, whether iron, steel, or nickel, was 

 on the whole the same for all. 



The dotted curve shown in Plate I. is the graph of No. VII. plotted in terms of 

 numbers proportional to the inductions in the tube instead of the fields. The abscissa 

 500 on the plate corresponds to an induction at the centre of about 7600. If we take 

 points at the same height on No. VII. graph and on the dotted curve, then multiply 

 the abscissa of the latter by 15*2, we shall get the induction corresponding to the field 

 which is given by the abscissa of the former. All the other graphs, if similarly treated, 

 would be transformed in much the same way, the early portion being greatly drawn 

 out. 



§ 9. The Bessemer Steel Tubes Nos. II. to VII. — For reasons already stated, the 

 results for Tube I., being quite untrustworthy, have been omitted. The general features 

 of the volume change are shown graphically in Plates III. and IV. It should be noted 

 that the scale to which the volume change is plotted is double that used in the preceding 



* This was proved by experiment. 



