In the same manner all the experiments might be taken and compared, and the law found true in every case. The discrepancies are comparatively small, and, as they appear to follow no law, are evidently errors of observation arising from unavoidable defects in the construction of the tubes and the varying rigidity in the plates of iron.

Two experiments made upon actual boiler flues fixed in their proper position in boilers show that there is no great departure from the same law up to 35 feet in length.

The longer of these two flues collapsed with 97 lbs. per square inch, whilst the shorter sustained 127 before giving

way.

II. Strength as affected by Diameter.

A precisely similar law is found to prevail in relation to the diameter. Tubes, similar in other respects, vary in strength inversely as their diameters. Testing this law in the same manner as the last, we have the following table:

6. Resistance to collapse of five-feet tubes.

The nearly constant numbers given in the last column, and representing the product of the diameter and collapsing pressure, or the collapsing pressure reduced to unity of diameter, verify the above law of strength.

7. Resistance to collapse of two feet six inch tubes.

The numbers in the last column are nearly constant.

III. Strength as affected by thickness of Plates.

It is found that the tubes vary in strength according to a certain power of the thickness, the index of which, taken from the mean of the experiments, is 2.19, or rather higher than the square.

Where L is in feet. Or for convenience of calculation,

With regard to cylindrical flues, the experiments indicate the necessity of an important modification of the ordinary mode of construction, in order to render them secure at the high pressures to which they are now almost constantly subjected. The weakness of flues on the present construction has already been shown. To remedy this defect, it is proposed that strong rigid rings of T, or angle iron, should be riveted at intervals along the flue, thus practically reducing its length, or in other words increasing its strength to uniformity with that of the exterior shell of the boiler. This modification, which is represented in Plate II. figs. 2 and 3, is so simple and yet so effective, that its adoption may be confidently recommended to the attention of those interested in the construction of boilers.

The following table of the proportions of boiler flues, supplementary to those given in the first series of Lectures on the proportions of the external shell, will be found worthy of attention.

TABLE showing the proportions of internal Boiler Flues, for resisting a collapsing pressure of 450 lbs. per square

inch.*

In the above table the length of the flue must be measured between the rigid supports. In an unsupported flue, as ordinarily constructed, the length is measured between the end plates of the boiler; in a flue as proposed above between the T iron ribs. For a collapsing pressure of 450 lbs. the safe working pressure would be 75 lbs. per square inch.

It will not be necessary to remark further on this subject, except to illustrate, by an example, the importance of these facts to the safety of the public. In the disastrous accident which attended the first trial trip of the Great Eastern the funnel of the boilers, which was surrounded by a water-jacket, gave way by collapse at what was probably a comparatively low pressure. This might easily have been prevented had the maker been aware

*This table was given in the third edition of the first seri Information."

X

extreme weakness of such flues when of large diameter and great length.

Fig. 70.

Fig. 70 shows the general arrangement of the boilers and funnel, covered by the jacket a a. The funnel, six feet in diameter, is in this case exposed to the pressure of the steam, together with that of a column of water nearly forty feet in depth, and these two forces were quite sufficient to collapse the funnel and cause the frightful explosion which occurred.

a

The great weakness of elliptical tubes, shown in the experiments on collapse, points out another source of danger in marine boilers, viz. the "take-up;" that is, the drawing in of the plates from a rectangular to a cylindrical form, where the flues join the funnel. This part requires the utmost attention, as it is not only the weakest part of the boiler from its form, unless very carefully stayed, but it is much exposed to overheating from the gases rising from the furnaces where it is above the level of the water, in which case its powers of resistance must be greatly diminished.

II. THE DENSITY OF STEAM.

If fatal accidents in the use of steam lead us to study the forms and proportions of boilers, the necessity for economy in its production and application should induce us to study minutely its properties. The knowledge of the latent heat of steam, the discovery of which had been made a short time before by Dr. Black, led Watt to his great invention of separate condensation, and since that time some of the most eminent scientific men have inves

tigated, theoretically and experimentally, its various properties. Much has, however, yet to be done; it is true we have the experiments of Robison, Southern, Ure, Dalton, Arago and Dulong, the Franklin Institute, and last, but not least, of Regnault in France, and many others of less importance than these, so that the relations of temperature and pressure, and the amounts of latent heat, total heat, and specific heat at all temperatures have now been determined with an accuracy which we can hardly hope to see excelled. But there are other properties of which we have hardly any experimental knowledge at all. To supply some of these defects I have been engaged upon a laborious series of experiments, in conjunction with my friend Mr. Thomas Tate, with a special view to determine the relations of temperature and density of saturated steam, and the laws of the expansion of superheated steam, which have not hitherto been made the subject of any reliable experiments. These experiments, not yet completed, have not been unattended with danger, from the necessity of employing glass tubes and globes at elevated temperatures and considerable pressure. Some of these tubes exploded, and on one occasion my assistant, when reading off the mercury levels, nearly lost an eye from the fragments. scattered about from a violent explosion. Every possible precaution has however been taken to obtain accurate results, and as in several respects these are new and interesting, a brief abstract may be given here in anticipation of the publication of the results in a detailed form.

For a perfect gas the law which regulates the relation between the temperature and volume, and known as GayLussac's or Dalton's law, combined with the law expressing the relation of pressure and volume, known as Boyle's or Mariotte's law, is expressed in the equation,