Similar experiments have been undertaken, to determine the comparative action of semi-cylindrical and parallelopipedal retorts.[11] The latter, when kept in action day and night, do not long retain their shape; their sides collapse, their capacity becomes diminished, their angular form causes the heat to act upon them unequally, in whatever manner it may be applied, in consequence of which they suffer more deterioration in some parts than in others. Besides, they require a much larger proportion of fuel for decomposing a certain quantity of coal than the cylindrical retorts. [11] At the Birmingham Gas Works. Semi-cylindrical retorts, with the base of the retort bent inwards, so as to give the vessel a kidney-shaped form, have likewise been tried. But this shape is still less advantageous; they could not be made to work uniform, they required more heat, and their deterioration was more rapid than cylindrical retorts. They could not be kept fit for use when worked day and night, more than about five months. And with regard to ellipsoidal retorts, it must be confessed, that the experiments that have as yet been made upon a large scale to ascertain their powers, are not of a nature to enable us to decide on their merits. No experiments have been carried on with retorts of this description in the metropolis for a sufficient length of time, with that care and attention which the subject demands, to ascertain their comparative power. From what however has been done, there is reason to believe that ellipsoidal retorts, might be found more advantageous, than those of a cylindrical form now in use. An ellipsoidal retort, 20 inches by 10 in diameter, and six feet long, weighs 14 Cwt. The reader will thus observe, that of all the forms of retorts which have been hitherto fairly tried, upon a large scale, it has been satisfactorily ascertained, (excepting only as to the ellipsoidal retorts), that the cylinder is the best form for decomposing coal in masses, from five to eight or ten inches in thickness. It is perhaps needless to state that in making experiments on the comparative value of the best form of cast-iron retorts, it is obvious that the operations should be continued for some months uninterruptedly; no conclusion can be drawn that may become practically useful in the large way, from processes carried on for a few weeks only. It is absolutely essential that the comparative trials be continued for months together, and that the inferences be taken from the total quantity of coal used during that period, compared with the total quantity of gas obtained, the deterioration of the retorts, and the time and labour expended.