great advances
have been made in the apparatus to be used. Here possibly, to a greater
extent than in any recording device, are the records of value in
determining relative evaporation, that is, an error is rather allowable
provided such an error be reasonably constant.

The apparatus for recording such evaporation is of two general classes:
Those measuring water before it is fed to the boiler and those measuring
the steam as it leaves. Of the first, the venturi meter is perhaps the
best known, though recently there has come into considerable vogue an
apparatus utilizing a weir notch for the measuring of such water. Both
methods are reasonably accurate and apparatus of this description has an
advantage over one measuring steam in that it may be calibrated much
more readily. Of the steam measuring devices, the one in most common use
is the steam flow meter. Provided the instruments are selected for a
proper flow, etc., they are of inestimable value in indicating the steam
consumption. Where such instruments are placed on the various engine
room lines, they will immediately indicate an excessive consumption for
any one of the units. With a steam flow meter placed on each boiler, it
is possible to fix relatively the amount produced by each boiler and,
considered in connection with some of the "check" records described
below, clearly indicate whether its portion of the total steam produced
is up to the standard set for the over-all boiler room efficiency.

Flue Gas Analysis--The value of a flue gas analysis as a measure of
furnace efficiency has already been indicated. There are on the market a
number of instruments by which a continuous record of the carbon dioxide
in the flue gases may be secured and in general the results so recorded
are accurate. The limitations of an analysis showing only CO_{2} and the
necessity of completing such an analysis with an Orsat, or like
apparatus, and in this way checking the automatic device, have already
been pointed out, but where such records are properly checked from time
to time and are used in conjunction with a record of flue temperatures,
the losses due to excess air or incomplete combustion and the like may
be directly compared for any period. Such records act as a means for
controlling excess air and also as a check on individual firemen.

Where the size of a plant will not warrant the purchase of an expensive
continuous CO_{2} recorder, it is advisable to make analyses of samp