of Large and Small Alternating Currents. 103 
the low P.D., 0°15 volt, required by No. 2, compare very 
favourably with other similar instruments. If No. 2 is 
shunted for 1000 amperes the loss will be only 150 watts, and 
a comparatively small shunt will be required. The same 
applies to No. 1, if used in series with a high resistance or 
connected to a transformer as a voltmeter. 
Temperature error—tIn the two present instruments this 
error is about 1 per cent. for 15° F’., the deflexion decreasing 
with increase of temperature. It is to be noted that this low 
value has not been obtained by making the moving-coil 
circuit have a resistance which is practically independent of 
temperature, but by so choosing the materials of the thermo- 
junctions that the thermoelectric power rises as the tempera- 
ture rises, and so partially balances the increase in resistance 
of the coil. I have found materials which will make this 
balance complete, and so do away with the temperature 
correction, but practical reasons which will be referred to 
later have weighed against their use. If the instrument is 
required only to measuresmall currents and not to beshunted— 
this is the case of a voltmeter for, say, 50 volts or over—then 
the temperature error can be easily reduced to a still smaller 
value by suitably choosing the alloy of which the heater is 
made. 
The iron-cased instrument No. 2 will stand a fairly large 
exeess-current without being injured. It will carry twice 
the normal current required to give a full-scale deflexion for 
half-an-hour without being permanently injured or its 
calibration altered. It has also had 24 times the normal 
current put through it for 20 seconds without injury. Of 
course, after such ill-usage, the calibration is temporarily 
altered a few per cent., but the instrument has so far always 
recovered itself in about + hour. 
The necessity of making the instrument to stand this large 
overload, limits very seriously the choice of materials and 
the design of the heater and thermo-junction. If I couid be 
certain that the instrument would not be seriously overrun, 
the present small temperature error could at once be got 
over or more sensitive instruments could be built. For 
practical work, whether on the switchboard or in the 
laboratory, it is essential that ammeters shall stand safely a 
considerable excess-current without being injured, for it is 
better to have a little less accuracy than no ammeter at all at 
the critical moment when the reading is required. 
I must now refer to an objection which is special to this 
type of instrument, namely, that its calibration is not exactly 
the same in the vertical as in the horizontal position. I think 
