mum demand.

We have made a number of experiments in Chicago to ascertain whether or not the Wright indicator could be used for this purpose. It has been our practice to sell alternating-current service on a sliding scale of prices-so much per kilowatt-hour-but we want to apply the maximum-demand system to this service. We made some experiments with the Wright indicator and came to the conclusion that we could not use it, for more than one reason. One reason is that the power-factor of the different phases is not the same, therefore we can not simply take the indication of one phase and assume that to be the maximum demand for the entire installation. Our contract department objected very seriously to the method of charging which involved charging for the idle current-that is, charging on the basis of kilovolt-amperes-because it put the customers' bills so high. We were forced to try to find some method of charging that would be fair to our alternating-current customers as well as to our direct-current customers, and we made an investigation of all our direct-current power business, taking about 2000 customers, involving about 10,000 horse-power, and found. the ratio between maximum load and connected load and applied these results to our alternating-current business. Of course that is approximately correct—that is, it is correct collectively—but the difficulty is that it is not correct in any individual case, so we still feel, nothwithstanding the fact that we are able to do business. on that basis, that we ought to have an instrument for determining the true energy.

Mr. Semenza points out very nicely, and Mr. Newbury pointed out the first day of the convention, the importance of the question of power-factor, and what the effect of low powerfactor was on the company's investment; that is, since the generators were heated up, lines heated up, transformers heated up, we should therefore charge for the idle current. But we must look at the matter also from a commercial standpoint, and not merely from an engineering standpoint. What we should do is to find a system that will measure the true energy, and then, perhaps, put our price somewhat higher, in order to compensate for the loss due to low power-factor. We have instruments that will take care of large installations, but what we want is an instrument that can be used for small installations; something not too expensive for general use.

MR. SANDS: Mr. Ferguson refers to the fact that the very large installations have a satisfactory instrument for ascertaining the true energy. I ask what that instrument is. I also ask what, in Mr. Ferguson's opinion, should be the proper lag for a threephase indicator. I am at present using the Wright indicator for ascertaining the demand. I find, however, that the curve of the present Wright indicator, to my mind, is altogether wrong. I do not believe that we have any right to impose the demand charge upon a customer when from 80 to 90 per cent of that demand can be registered in the sharp peak of a load lasting for only a matter of five or six minutes. If we are going to surround the sale of electric power with a lot of theoretical conditions we had better get out of business. We have to compete with the isolated plant and with the man who has his own steam plant. I have served my purpose by forgetting all about power-factor and simply taking 75 per cent of the volt-ampere reading of the demand indicators and charging for the demand at the primary and for the remainder at the secondary rate. I should like to ascertain what would be, in Mr. Ferguson's opinion, the proper lag if a proper three-phase indicator could be found.

MR. JUNKERSFELD: Two things of primary importance are brought out in this paper; one is that the load-factor should be taken into consideration, which is a question that this convention has discussed a great many times; the second point is the matter of power-factor.

As has been pointed out, the production of energy is in no sense dependent on the power-factor. The distribution of energy is very largely dependent on the power-factor. Commercial considerations, however, come in, which complicates the explanation of this effect of the power-factor to the customer, either when soliciting his business or later when explaining his bill.

The polyphase circuits are nearly always unbalanced, and for that reason, as Mr. Ferguson indicated, the ordinary Wright demand indicator gets us into difficulties at once. There may be a very large percentage of difference of current in the three legs of the system. We have come to the conclusion in Chicago that perhaps the simplest way is, as one gentleman has just said, to forget the power-factor at the time, but not to forget it at the time of fixing the rate; doing so in the interest of simplicity.

The integrating wattmeter is an instrument developed to a

high standard. The art of clock-making is also developed to a high standard. With a combination of these two we are handling a great many large customers by simply using the integrated output for a definite period of time-one hour, half an hour, or fifteen minutes, as the case may be. In supplying large railroad systems, where there are hundreds of cars and the load curve shows no sharp peaks, the average demand for one hour does very well; but for industrial power or for interurban railways, where the peaks are sharp, a shorter interval of time should be taken. In these cases it would be fairer and more equitable to both company and consumer to use the average demand over a fifteenminute, or even shorter, period. That can be done by the timeclock arrangement to the meter. We have used such an arrangement for a year, and in this way we get the average demand for a period that may be agreed upon by the contract and justified by the conditions. In that way you get something that the average customer can understand when you are soliciting business, and it is a simple thing to explain. In this kind of power it costs more for distribution if the customer's demand is at the low power-factor, but the thing to do is to keep that in mind when fixing the price.

MR. Dow: The question of a demand indicator for polyphase circuits is now being studied, not only in the United States, but in England and on the Continent, and the General Electric Company has, to my knowledge, done a good deal of experimental work on it. The downright difficulty, however, is not to combine the clock suggested by Mr. Junkersfeld with the integrating meter, nor to devise an instrument to give a record of the load at any time or at any number of times, nor to integrate for any period. The primary difficulty is to determine just what we want to find out, and these expressions of opinion by Mr. Ferguson are very much in line with the request of the meter manufacturers to-day, that the people who want to buy such meters get together and settle definitely what kind of meter it is that they require; and, having settled that, what price they are willing to pay for it. Practically, you can get anything. It may become an exceedingly complex instrument; and while the complexity is warranted in such instances as Mr. Junkersfeld talked about, the cost of such an instrument would not be warranted in the average

small polyphase motor installation. What, then, is wanted, gentlemen, and what I hope will come as a consequence of this discussion, is a definite expression of opinion as to what one device is wanted, what the device shall do; or what two alternative devices are wanted; and that thereafter consideration be given as to what the selling price should be to warrant its manufacture in quantity and its purchase by our members. In making this statement I know I am expressing the very recent views of the meter manufacturers.

MR. PHILIP TORCHIO (New York city): I desire to confirm what Mr. Dow has said. You can get from the manufacturers at the present time any instrument you wish for measuring large amounts of power. Some of the instruments that are in operation around Niagara Falls register the maximum output in any one hour; other instruments register the maximum demand over five minutes. Some of the instruments are based on the dial, or ticker, system; others are recording instruments, which draw the curve of the load diagram for 24 hours, or for weeks or months. All these instruments are being manufactured, and can be secured to meet any particular purpose.

Referring to the subject of power-factor, I think we who supply light from our central stations are very much interested in good power-factors, not only for the reason that we do not want to overload our cables, transformers and generators, but also for keeping the regulation good; and with low power-factor and fluctuating loads the regulation, of course, becomes very bad. The manufacturers ought to be impressed with the fact that induction motors must be made with the best power-factor that can be had, and the controlling and starting devices be made to reduce the inrush current to a minimum. In an instance where we were asked to bid on very large motors to supply water-pumping stations, operated by motors of about 800 horse-power each, it was made possible to start the pumps under full head and still not have an inrush of current of more than 50 per cent over full-load current for a few seconds.

MR. R. W. ROLLINS (Hartford, Conn.): Mr. Ferguson made the statement that instead of installing Wright demand. meters on his polyphase system they were basing their calculations on the results of their experiences on the direct-current

system, based on 10,000 horse-power. It would be interesting to me to ascertain the relation of the maximum demand to the full 10,000-hp rating.

MR. FERGUSON: My object in bringing this matter up was not to discuss instruments and details of instruments, but to find out whether or not there was a necessity for such an instrument. We have discovered that there is a necessity for it. I assumed that it was generally known that the ordinary integrating wattmeter could be used for large power installations. That was not the point to be discussed. It was to find out whether or not there was a necessity for an inexpensive meter that could be applied to all of our power business.

In large railway or industrial power installations the matter of $50 or $75 or $100 for a meter does not cut any figure at all, and such a meter is on the market; or even the ordinary integrating wattmeter can be used by taking readings at intervals of an hour, 30 minutes, or 15 minutes, or any time that may seem best adapted to suit the particular conditions. I think that will answer Mr. Sands' question as to what kind of instrument I refer to.

I will answer Mr. Rollins' question, and think it will also answer Mr. Sands' question regarding the percentages used. We made a study of 10,000 horse-power, covering a total of 2000 motors, on our alternating-current system, and found that the ratio between the maximum load and the connected load was 64 per cent, taking the average of all this business. But the percentage varies a little under different conditions of installation; where a customer has one motor the ratio between maximum load and connected load is higher than where there is more than one motor installed, therefore you have to differentiate between. the two. You can see why that is: If a man has fifty 1-hp. motors they will not all be used at once, whereas if he has only one 50-hp motor he is using it all the time. We found it was better to make the classification according to horse-power. This tends to take into consideration approximately all the difference between groups of motors and installations of one motor. I will read the classification under which we are working:

Class A-Installations up to 10 horse-power, where one motor is used, we use a figure of 85 per cent of the connected load; where more than one motor is used, 75 per cent of the connected load.