hich, in
the simple trials, rose from 11.7 per cent. at 40 lb. absolute
pressure to nearly 30 per cent. at 110 lb. and at 90 lb. was 24.8 per
cent., was at 90 lb. only 5 per cent. in the compound trials. In the
latter, at 160 lb., it increased to 17 per cent., but, on repeating
the trial with triple expansion, it fell to 5.46 per cent. or to 4.43
per cent. in another trial not included in the table.

On the other hand, from the greater loss in passages, etc., the
compound engine must always give a smaller diagram, considered with
reference to the steam present at cut-off, than a simple engine, and a
triple a smaller diagram than a compound engine. Nevertheless, even at
80 lb. absolute pressure, the compound engine had considerable
advantage, not only from lessened initial condensation, but from
smaller loss from clearances, and from reducing both the amount of
leakage and the loss resulting from it. These gains became more
apparent with increasing wear. The greater surface in a compound
engine had not the injurious effect sometimes attributed to it, and
the author showed how much less the theoretical diagram was reduced by
the two small areas taken out of it in a compound engine than by the
single large area abstracted in a simple engine. The trials completely
confirmed the view that the compound engine owed its superiority to
reduced range of temperature. At the unavoidably restricted pressures
of the triple trials, the losses due to the new set of passages, etc.,
almost neutralized the saving in initial condensation, but with
increased pressure--say to 200 lb. absolute--there would evidently be
considerable economy. The figures of these trials showed that the loss
of pressure due to passages was far greater with high than with low
pressure steam, and that pipes and passages should be proportioned
with reference to the weight of steam passing, and not for a
particular velocity merely.

The author described a series of calorimetric tests upon a large scale
(usually with over two tons of water), the results of which were
stated to be very consistent. After comparing the dates of initial
condensation in cases where the density of steam, the area of exposed
surface, and the range of temperature were all variables, with other
cases (1) where the density was constant and (2) where the surface was
constant, the author concluded that, at four hundred revolutions per
minute, the amount of initial condensation depended chiefly