to the Longitudinal Expansion in Rods of Spring Steel. 31 



Fig. 3, shows part of it. One of its axes is parallel to the wall, 

 the other is at right angles thereto ; I shall call the first the 

 f axis, and the second the 77 axis. 



The boards B, B, and C support the brackets b, b, and c, 

 which project somewhat further than the arms a, a, and of which 

 the first two support two telescopes /3, ft, magnifying about thirty- 

 times. The visual axes of these are placed vertically. To effect 

 this, a cross consisting of two wires is stretched below each 

 object-glass, and the telescope is so arranged that the image 

 which a mercury horizon gives of the intersection of these wires 

 coincides with the cross-hair of the telescope. 



On the bracket c is suspended the rod whose changes of form 

 are to be measured. Through the bracket, and near its front end, 

 a vertical rectangular aperture is made, which is partially filled 

 with a piece of wood ; this is supported by the four lateral screws 

 7, 7, 7', 7', the points of which catch in cavities made in the 

 wood. Of these four screws, the first two are in a horizontal, 

 and the last two in a vertical plane. By their means the piece 

 of wood can be shifted in the direction of the f axis, and rotated 

 about two axes, one of which is almost vertical, and the other 

 almost parallel to the 77 axis. The piece of wood is perforated 

 vertically; through the perforation the tail of a small vice is 

 passed from below, and, by means of a nut, fastened against the 

 piece of wood, so that it only moves with considerable friction. 

 In the jaws of the vice a small piece of steel plate is held, which 

 has an aperture just as large as the section of the rod under 

 investigation- Through this aperture the rod is pushed to its 

 middle, and soldered here by a little tin. By means of these 

 arrangements a position can be given to the rod (which is indi- 

 cated by d, fig. 2, Plate I.) in which its axis is horizontal and 

 parallel to the f axis, and the cross rods e, e fixed to it are as 

 nearly horizontal as possible, even if they are not quite parallel 

 to one another. For this purpose a level provided with hooks 

 is so suspended to the rod d, that its middle is under the middle 

 of the rod, and by turning the screws <y ! , 7' its bubble is brought 

 to the centre. Thereupon the level is suspended to one of the 

 cross rods e, e, and this made horizontal by turning the piece of 

 metal which holds the rod d. If both cross rods are parallel to 

 one another, the second must also be horizontal. It is ascer- 

 tained whether this is the case by suspending the level to it. A 

 small deflection need not be taken into account. If it occurs, 

 the rod d is most conveniently arranged so that both cross rods 

 deviate from the horizontal to the same extent in opposite direc- 

 tions. Finally, the motion which is necessary to arrange the axis 

 of the bar d parallel to the £ axis, can be produced partly by 

 turning the vice in the piece of wood which supports this, and 



