680 REPORT—1904. 
energy loss being measured by a wattmeter. The hysteresis curves of this figure 
are based on a single test by a Ewing tester, for each sample of iron, worked out 
for 50~ and 100~ by comparisons with standards tested at 4,000 B, 100~ by 
ballistic galvanometer, the values at other B’s being calculated on the Steinmetz 
B'¢ ratio. The results need but little explanation to show the marked increase 
of loss with thickness. 
These tests do not lend any support to the common belief that with ordinary 
degrees of lamination the eddy-current loss is reduced to negligible proportions. 
On the contrary, they show that the eddies loss is of the same order of importance 
as that due to hysteresis, 
In calculating losses in iron for hysteresis tests, it is usual to assume that the 
rate of increase of eddy loss is proportional to the square of the thickness, the 
square of B and the square of the periodicity. An examination of the curves 
will show that in some respects these assumptions are not confirmed, thus :— 
Thickness.—As between ‘0136 and ‘0189 inch, the eddy loss on the average is 
practically proportional to (thickness)?, but as between ‘0136 and ‘0254 inch the 
average increase is about 30 per cent. less than the ratio of (thickness)*, the 
increase being from 1 to about 2°4 instead of 1 : 3:5. 
Eddies and g.—On the average the increase is substantially in accordance 
with the B* assumption. 
Eddies and Periods.—The increase is rather less than the usual (periods)? 
assumption, being about 1 : 3-4 instead of 1 : 4. 
The departure from the (thickness)? and (periods)’ ratios of increase may 
reasonably be explained by supposing the eddy circuits to have self-induction. 
The six total loss curves will be found to be rather steeper than Bl curves. 
It often happens, however, that the curves of total loss are substantially BY 
curves, showing that in such cases the eddy constituent of the loss as well as the 
hysteresis closely follows the Steinmetz ratio. 
Fig. 2 illustrates this. It gives the total loss, measured by wattmeter, of 
iron ‘014 inch thick at 50 periods and at 100 periods and at various B’s. The 
wattmeter readings are shown by round points, the points marked + being Bl® 
values. 
This iron is representative of good transformer iron, now obtainable in quan- 
tities in England. Slightly better iron is sometimes got, but not often nor in 
large quantities. 
Fig. 5 shows the result of total loss test by wattmeter of some ‘0124 inch iron 
at two temperatures—viz., 40° F. and 167° F., together with a ballistic galvano- 
meter test for hysteresis of the same iron taken at 2,000, 5,000, and 8,000 B. 
The three hysteresis values so obtained fall on a Bl curve. 
It will be found that the total loss curves (at the two very different tempera- 
tures mentioned) also follow B!° curves, which are shown thus +, the wattmeter 
readings being shown by the round points. 
Although the total loss generally closely approximates to B'% curves, it may be 
either less or more steep than such curves. Fig. 4 shows examples of somewhat 
greater steepness; figs. 2 and 5 show coincidence with B!® curves; and the authors 
have also found curves less steep. Examples of the latter may be seen, eg., in 
the total loss tests given in Kapp’s ‘Transformers’ (1896, p. 23) of iron carried up 
to nearly 7,000 B, which will be found on examination to be less steep than B® 
curves. The table of hysteresis values (going up to 15,000 g) given at p. 108 of 
Ewing’s ‘ Magnetic Induction in Iron’ (8rd edition) will also, on examination, be 
found to form a curve perceptibly less steep than BY*. As on the whole, how- 
ever, the evidence is in favour of the Steinmetz ratio for the hysteresis (though 
by no means invariably so), it is probable that the variations in the forms of 
the total loss curves are due to differences in the ratio of increase of the eddy 
currents. The various results point to the need for making wattmeter tests of 
total loss in iron under working conditions, and that it is not safe to rely only on 
hysteresis tests, 
