Moreover, many of these tests are nowadays, in a well-equipped testing-house, carried out in conjunction with the use of heat. It stands to reason that a part of a machine which will have to work under considerable heat may have to be of different material from a part which works under a normal temperature. In some cases the bar is surrounded by a spiral wire through which electric current is passing, and by the regulation of this current any desired temperature can be set up in the bar. Or it may be placed in a bath of hot oil in such a way that the bar shall be raised to any temperature required, without interfering with the machinery which exerts the tension or pressure, or whatever it be. Years ago such elaborate tests as these were never thought of. There are certain well-known figures, to be found in all engineering text-books, which give what stresses different materials ought to be able to stand, and these were, and are still, to a large extent, relied upon, it being taken for granted that the material used will be up to the average standard. In large and important works, however, the testing has been developed upon scientific lines, so that it is known from actual experiment what each particular thing is capable of. This not only means security but economy, for it is sometimes found that a substance is stronger than it is thought to be, and so things made of it can be designed to give the requisite strength lighter and cheaper than they would have been otherwise. Some of the machines employed are of enormous strength, capable of exerting a pull or a compression of, it may be, 100 tons or more. They are often made, too, with self-recording appliances, whereby the course of the test is set down automatically upon a chart. For example, when a bar is being tested for tension, it is desirable to know not only the actual pull under which it came in two, but the behaviour of the test piece during the period before that. It begins to stretch as soon as the tension is applied, theoretically at all events, and if the metal were perfectly ductile it would stretch continuously as the load increases, until at last the breaking stress is reached. But in actual practice it probably stretches somewhat by fits and starts, and a record of that fact will be of great value in estimating the strength of the material in actual work. For such, an automatically made record, which can be studied at leisure, is of the utmost importance. But perhaps the finest instance of scientific methods in manufacture is to be found in the methods by which standard parts of machines are measured, so as to ensure that they shall be interchangeable.