180 PROCEEDINGS OF THE AMERICAN ACADEMY. 



alloy before we fully realized the great difference in ductility between this 

 and pure platinum. Because of its extreme ductility gold was then tried. 

 This worked beautifully at first, but finally failed because the 22-carat 

 gold solder employed in making the tube disintegrated under the action 

 of the hot water. Recourse was then had to pure platinum, which com- 

 pletely solved the difficulty. The platinum tube is first expanded by 

 drivinc in some plugs of cotton with the help of a hammer and a brass 

 rod almost as large as the inside of the tube. Seamless tubing might be 

 advantageously used here, but we used a tube made of sheet platinum 

 soldered with pure gold, and this proved to be entirely satisfactory. 



The small tube, T 1? is conveniently made by rolling up tightly some 

 thin sheet platinum and then flowing gold in to fill the spaces between 

 the convolutions. This gold is fed in from the outside, while the whole 

 tube is kept hot in a large blast-lamp flame. Care must be taken not to 

 use too much gold ; otherwise a drop may form inside the tube, and its 

 removal by drilling is extremely difficult. Owing to capillary forces, 

 no gold will go to the space iuside until the smaller spaces between the 

 convolutions are all filled, so that there is no danger so long as too much 

 gold is not employed. In this, as in all other operations when gold is 

 used in soldering platinum, the piece should be kept hot no longer than 

 is absolutely necessary, because the gold rapidly alloys with the platinum, 

 and the resulting alloy is more crystalline in structure than either of the 

 constituents and has not their ductility. To make the joint between T t 

 and the lining of the cover stronger, the tube is reinforced above this 

 point, as shown by the drawing. This was necessary in our earlier 

 apparatus before the lining was screwed down to the cover, but is 

 probably not necessary in the later form. 



As the upper end of the tube T x is to act as a valve seat, and as 

 therefore there will be a good deal of downward pressure at this point, 

 the tube has to be well expanded into the conical cavity in the steel at 

 V (Fig. 1). Because of this, and of the further fact that the valve seat 

 should be as soft as possible, it is better to make the upper end of T x of 

 solid gold, boring it out later. This is done as follows : The tube is 

 first packed full of asbestos, to prevent gold from getting into it. A 

 band of thin platinum foil is next wound tightly around the upper end 

 and bound on by means of a platinum wire. This band is then pushed 

 partly off of the end of the tube, so as to make a small projecting tube ; 

 and pure gold is melted into this until it is full. The platinum foil on 

 the outside of the gold is now filed off. Both tubes are now attached to 

 the cover lining and inserted in place in the cover. Holding the lower 



