In many cases one can resort to the device already mentioned of enclosing a rubber or tape-wrapped joint between two tubes in a bath of mercury, but in this case the glass must be clean and hot and the mercury also warm, dry, and pure when the joint is put together, otherwise an appreciable air film is left against the glass, and this may creep into the joint. Perhaps the easiest way of making such a joint is to use an outer tube of thin clean glass, and bore a narrow hole into it from one side to admit the mercury; if the mercury is to be heated in vacuo, it is better to seal on a side joint. It is always better, if possible, to boil the mercury in situ, which involves making the wrapping of asbestos, but, after all, we come back to the position I began by taking up, viz. that the easiest and most reliable method is by fusion of the glass--all the rest are unsuitable for work of real precision. I should be ungrateful, however, were I not to devote a few lines to the great convenience and merit of so-called "centering cement." This substance has two or three very valuable properties. It is very tough and strong in itself, and it remains plastic on cooling for some time before it really sets. If for any reason a small tube has to be cemented into a larger one, which is a good deal larger, so that an appreciable mass of cement is necessary, and particularly if the joint requires to have great mechanical strength, this cement is invaluable. I have even used a plug of it instead of a cork for making the joint between a gas delivery tube and a calcium chloride tower. (Why are these affairs made with such abominable tubulures?) The joint in question has never allowed the tube to sag though it projects horizontally to a distance of 6 inches, and has had to withstand nearly two years of Sydney temperature. The cement consists of a mixture of shellac and 10 per cent of oil of cassia. The shellac is first melted in an iron ladle, and the oil of cassia quickly added and stirred in, to an extent of about 10 per cent, but the exact proportions are not of importance. Great care must be taken not to overheat the shellac. APPENDIX TO CHAPTER I ON THE PREPARATION OF VACUUM TUBES FOR THE PRODUCTION OF PROFESSOR ROENTGEN'S RADIATION [Footnote: Written in May 1896.] WHEN Professor Roentgen's discovery was first announced at the end of 1895 much difficulty was experienced in obtaining radiation of the requisite intensity for the repetition of his experiments. The following notes on the production of vacuum tubes of the required quality may therefore be of use to those who desire to prepare their own apparatus. It appears that flint glass is much more opaque to Roentgen's radiation than soda glass, and consequently the vacuum tubes require to be prepared from the latter material. Fig. 39.