94 Sir Charles A. Parsons [March 10, 



up to its speed for maximum efficiency. In large marine work, 

 where weight and space are of importance, K varies from 80,000 to 

 120,000 or even to 140,000. With K = 80,000, a loss of efficiency 

 of about 9 per cent, below the highest attainal)le is accepted. With 

 K = 120,000, the deficit is only about 1^ per cent. 



There are many forms of turbines now on the market, but we 

 need only consider four chief types, which are : — 



First, the compound reaction turbine, with which we have been 

 dealing, representing more than 90 per cent, of all marine turbines in 

 use in the world, and about half the land turbines driving dynamos. 



Secondly, the de Laval, which is only used for small powers. 



Thirdly, the " multiple impulse compounded," or Curtis, which 

 has been chiefly used on land, but which has been fitted in a few 

 ships. 



Lastly, the compound reaction type, with one or more " multiple 

 impulse elements " added to replace the reaction blading at the high- 

 pressure end. 



We may dismiss the numerous other types as simply modifications 

 of the original type, without any scientific interest. 



Let me explain the latter types. The multiple impulse principle 

 is the only substantial innovation since the compound reaction and 

 the de Laval turbines came into use. It was proposed l)y Pilbrow in 

 1842, and first brought into successful operation by Curtis in 189(). 

 A little consideration should be given to it as involving some 

 characteristic points of difference from what has been said about 

 reaction lilading. It will be seen that Curtis used the de Laval 

 divergent nozzle, and that he also uses compounding, but generally 

 only 5 to 9 stages as compared with 50 to 100 in the compound type. 

 The same principles as regards velocity ratio apply, but owing to the 

 repeated transfer of the steam between fixed and moving buckets at 

 each velocity-compounded stage, the liest velocity ratio in a four-row 

 multiple impulse is only one-seventh, and the best obtainable efficiency 

 44 per cent., and therefore much lower than reaction 1 (lading under 

 favourable conditions. 



The good points of the multiple impulse type are that there is 

 very little loss by leakage, and that therefore, in spite of its low 

 efficiency, one or more nmltiple impulse wheels can in certain cases 

 usefully replace reaction blading at the entry to the turbine, because 

 in slow revolution turbines of moderate power the blades are short at 

 the commencement, and there is consequently much loss by leakage 

 through the clearance space. As a rule, one nndtiple ini])ulse wheel is 

 generally preferred, and is followed by reaction blading ; the expansion 

 ratio on to the wheel is about threefold, and it generates about one 

 quarter of the whole power. Occasionally several wheels in separate 

 chambers are placed in front of the reaction blading, but there are 

 serious practical drawbacks to this arrangement. The nudtii)le im- 

 pulse wheel at the commencement has a further advantage in that, 



