36 



THE CIVIL ENGINEER AND ARCHITECT'S JOURNAL. 



LJanuary, 



worked, as well as the a.ptatinp apparatus, by the descent of water 

 flowiiifr from the loiiff le^ of the syiihon, which jrives motion to a 

 small water-wlieel in connection witli tlie bevel wheel, f;:earintr, and 

 band ; or they may be worked by a steam eng-ine or other ))ower. 

 All tlie condensable products ccdlected in the asritatinif chamber 

 and refriiieratinif pipe <j. How throutrh the ]dpe /i, into the chaml)er 

 f and throujjh the opening r, at the level of the dotted line, into a 

 receiver. 



Fig. 9 is a plan, and fig. 10 a sectional elevation, of the " am- 

 moniacal filtering towers," steam chambers, and condensers, com- 

 bined ia one apparatus. The gas takes the course indicated by the 

 arrows in the towers 1, 2, and 3, entering each at tlie bottom and 

 out at the top, and thence into the steam chambers 4, 5, 6, under- 

 going the steaming ;md condensing before explained ; «, o, steam 

 pipes, H-ith cocks to regulate the steam; h, the entrance steam pipe 

 from the boiler ; c, c, c, separate condensers, with the entrance and 

 e.xit i)ipes ; rf, the tank for the ammoniacal liquor, pumped up 

 through the pipe i; ; the tank has two divisional plates/, /; iixed to 

 the top and sides, and descending to within a few ijiclies of the 

 bottom of tlie tanl\, and is sealed at the \e\e\ of the dotted line by 

 the liquid ammonia. 



To insure the gas flowing from one tower to the other, each has 

 a pipe, !)-, connected with the tank and rising in it to the height of 

 the dotted line, at wliich level the ammonia flows through the pipe 

 o, into its particular tower. 



Instead of the arrangement of the filtering towers, several per- 

 forated divisional plates, n. as shown in fig. II, may be adopted, 

 the gas flowing from the tower into the chamber tlirough the pipe 

 711, in order finally to escape at the pipe y. 



REVixrws. 



All Essay on the Air-pump and Atmosplierio Railway ; containing 

 formultf find rules for ca/cidatiny the various quantities contained in 

 Mr. R. Stephenson' s report on atmospheric propulsion, for the Direc- 

 tors of tlie Chester and Holyhead Railway Company. Uy A^^illiam 

 TuENBiLiy, author of a treatise " On the Strength of Cast-Iron," 

 &c. London : AMlliams. 1847. 12mo. pp. 96. 



The object of this excellent little treatise is a general exposition 

 of the theoretical principles of atmospheric railways. That the 

 leakage of the main tubes of tliese railways involves a loss of 

 power, is obvious to every one in tlie slightest degree acquainted 

 with the subject ; hut it requires much more than superficial know- 

 ledge to estimate the precise amount of loss corresponding to a 

 given rate of leakage. Mr. TurnbuU has addressed himself very 

 successfully to the task of substituting exact principles for general 

 notions respecting the mechanical defects of atmospheric pro- 

 pulsion. 



The first part of this work comprises a history of the air-pump, 

 and demonstrations of several known formidaj by which its effects 

 are estimated. In the second part, these formul* are applied in 

 detail to the case of the Kingstown and Dalkey Railway. Not- 

 withstanding the imperfect success of the method of substituting 

 stationary air-pumps for locomotive engines, the subject is one of 

 permanent interest to the engineer, on account of the number of 

 beautiful scientific and mechanical problems which it presents to 

 his attention. Ccmsidered merely as an instructive exercise, the 

 theory of atmospheric propulsion deserves to be thoroughly mas- 

 tered by every student of practical science. It is this consideration 

 which indu(ves fis to give a brief sketch of Mr. Turnbull's method 

 of inrve^itigatiiui. j ', ■, 



Wlieji a t^-aili •o_n/tj>e atmospheric railway lias attained its uni- 

 form velocit^y, it is oljvio'is that, if there were no leakatje, the 

 pump-jiistoit and the train-piston must both describe tlie same 

 space in a given time- — thfiA is, the void made by the one in a given 

 time must be filled up by the otiier. For example, if the relative 

 diameters of the main tube and pump were such, that ten feet of 

 tjie lengtli of the former liad the same cubic capacity as one foot 

 of the length of the lattfr, the train-piston would travel ten feet 

 while the [mmp-pistou travelled one. Otherwise, if tlie puinp- 

 fTiston tra\telled at ^ greater relative velocity, the degree of 

 va«Tj.iirti iKiiihl be raised, and the train accelerated; if the pump- 

 pistfoii t^-iiyeUcd Bl a smaller rehitive velocity, the degree of 

 vacuum \ViJuld be diminisheil, and tlie train retarded : and either 

 case is contrary to the hypothesis of uniform velocity of the train. 



The exact relation, however, between the uniform velocities of 

 the two pistons only obtains on the hypothesis that there is uo 



b 



leakage. The principal problem is to ascertain the modification 

 due to that detect ot tlie apparatus. Tlie requisite data for this 

 investigation are obtained by the following experiment : — After 

 the tube has been exhausted to a certain extent, the whole ap- 

 paratus is suffered to remain quiescent, no train being dispatched. 

 The leakage will then go on till the equilibrium of the air inside 

 and outside tube be restored. By observing the rate at which the 

 barometer-guage falls during the interval, we get — not the rate of 

 leakage — but data from which that rate may be calculated. 



The density of air is proportional to the weight, and therefore 

 height, of the column of mercury. Take 30 inches as the 

 height of mercury ciu-responding to tlie atmospheric pressure ; 

 then, if the barometer-guage of the exhausted tube show, for the 

 pressure in it, a height equivalent to 10 inches of mercury (for 

 example), the density in the tube would be to that of the external 

 air as ll) : 30, or would be ^rd the ordinary density of air. If, 

 after the leakage has gone on some time, the barometer-guage 

 show a lieight equivalent to 20 inches for the pressure in tlie tube, 

 the density will be ^, or |rds that of common air. Tlie difference 

 between the densities in the tube at the two respective periods is 

 frds-i^rd (=^rd) that of common air. Consequently, if the 

 quantity of air which has entered the tube in the interval, be 

 supposed to have diffused itself equably throughout the tube, that 

 quantity is equivalent to the tube full of air at a density ^rd that 

 of common ab, or, which is obviously the same thing, one-third 

 the tulje full of common air. This reasoning applies generally, 

 and gives this simple rule — that the cubic quantity of air admitted 

 by leakage during any interval, is equal to tlie cubic capacity of 

 the tube multiplied liy the fraction expressing the difference of 

 densities during that interval. (The barometer-guage is so gra- 

 duated, that for the words, "fraction e.xpressiug the difference of 

 densities" in the above rule, we may substitute, " difference of 

 gauge-heights divided by 30.") 



If this quantity of air were divided by the number of minutes 

 of the interval, the result would be the rate of influx per minute, 

 supposing that rate uniform. This method of inxestigation is, 

 however, liable to an objection, which our author well states as fol- 

 lows : — 



" We have calculated for the extreme indications of the vacuum gauge, 

 and divided by tlie number of minutes that elapsed during the observatiou, 

 for the average leakage per minute. Now this method would be perfectly 

 just, on the supposition that the quantity of leakage is constant, or of tlie 

 same amount in equal times ; but the idea of a constant amount of leakage 

 is altogether incompatible with what we know to take place, when air of 

 atmospheric density is allowed to Sow into a vessel containing air of a less 

 density. Here it is obvious that the air in the vessel is continually ap- 

 proaching to a state of equilibrium with that without, and consequently the 

 velocity of influx is continually diminishing until the equdibrium obtains." 



He then proceeds to show, that in those experiments on the 

 connecting pipe of the Dalkey line, in which the heights of the 

 gauge were taken every minute, though the successive differences 

 of those heights for successive minutes were nearly equal, they do 

 not indicate a uniform rate of leakage, but lead to the directly 

 opposite conclusion, that the leakage was far more rapid at the 

 beginning of the experiment than at its conclusion : and he then 

 makes the following important remark in reference to Mr. Ste- 

 phenson's report : — " We are somewhat apprehensive that, by assuming 

 a constant amount of leakage for the connecting pipe, some very er- 

 roneous deductions must have been made." 



" But with regard to the valve tube the case is very different ; for it is 

 easy to conceive that, as the longitudinal slot or aperture is covered with a 

 flexible substance, this substance will readily accommodate itself to the 

 pressure as the exhaustion goes on, and by thus diminishing the area of the 

 aperture as the velocity of influx increases, a constant amount of leakage, 

 or nearly so, may happen to be maintained : at all events, it is not incon- 

 sistent with the maxims of accurate science, to admit that such may be the 

 case, and it actually appears from experiment that the supposition is not far 

 from the truth." 



If it be conceded that the leakage of the connecting pipe is an 

 avoidable evil, and may therefore be assumed to be wholly reme- 

 died, we have very simple means of calculating the effect which 

 the leakage of the main tube has on the velocity of the train. As 

 the assumption of uniform leakage in this tube is somewhat dan- 

 gerous, let the leakage corresponding to any proposed- working 

 vacuum lie ascertained by a separate experiment « ith the barome- 

 ter-gauge. We have explained how to calculate, from the fall of 

 the gauge, the ([uantity of external air which enters the tube per 

 minute. It may be calculated liy very simple arithmetic what 

 length of tube this quantity of air would by itself occu]>y, if dilated 

 to the supjiosed working density. And that length of tube is the 

 measure of the loss of speed of the train during the minute ; for 



