policy is to make public our various schedules of rates, and I think that as the result of these claims we shall have to be even more careful to see that every customer of the company knows every schedule offered by the company, inasmuch as it has been claimed -although, we think, unreasonably-that customers have not had an opportunity of knowing and understanding these various. schedules. Having done that-and, of course, specifically offering a customer another schedule whenever it shall come to our notice that he would be benefited by it—we feel that in the event of a case arising where the company has not intentionally required the customer to pay more under one schedule than he would under another, the customer will have no claim for rebate because he happens to have paid more money under one schedule than he would have done had he known in advance of the making of the contract all that he did at its conclusion. I think if that question should be determined definitely, the issue as to our rights to have differential rates would not be a bother to any of us in our respective territories.

MR. GEISER: This discussion has opened up a question that has bothered the central-station people in their endeavor to make equitable, uniform and profitable rates-that is, how to differentiate between different customers. If the company is legally allowed to make different rates to different customers under different conditions, how shall that be accomplished so that all customers can get the right rates? The question in the Brooklyn case was the changing from one rate to another. Is it not possible to adopt a system or schedule of rates that will take care of that with one schedule? We have been laboring along that line with a fair degree of success. We have in Waynesboro one scale of rates for all customers, from one light to 1000 lights, from $1 a month to $600 or $800 a month. When a customer grows out of one class and into another-which appears to have made the trouble in Brooklyn-he gets the benefit of it without any special agreement. It is not entirely on the plan of charging a certain rate. for the first hour's use and another rate for the second hour's use per day, but it provides for something that method does not. provide for. Our rate goes further, and provides for the large as well as for the small customer, at the same time differentiating between the long-hour and the short-hour consumer. Instead

of using one hour per day or two hours per day as the basis of the sliding scale, what we call a "service unit" is used as the basis. That service unit is composed of two factors: first, a certain number of hours' use, and second, a definite fixed amount in kilowatt-hours, applied to all customers alike. For instance, five hours' use of the capacity demand, plus six kilowatt-hours, is the service unit or basis of our sliding scale. The first five hours' use plus six kilowatt-hours, or the first service unit in our rate, is charged at the initial rate-the maximum rate; the second five hours plus six kilowatt-hours, or the second service, at the second rate, and another five hours plus six kilowatt-hours' consumption, at the third rate, and so on. We find that this method takes care of all classes of customers, with the curve of rates very closely approximating our cost curve for the different classes of service. The shape of this rate curve can be varied, of course, by changing the figures for these various points, the fixed amount, the number of hours' use, and the number of service units to be charged at a certain rate. As an illustration, our rate is sixteen and two-thirds cents per kilowatt-hour for the first service unit or the first five hours' use of the capacity plus six kilowatts. The second service unit is charged at 15 cents; the third at 10 cents; the next 10 service units at three cents, and all beyond 13 service units at one cent per kilowatt-hour. One cent is pretty low, but it applies only to a portion of the consumption after all the fixed costs have been taken care of by the higher rates, and any additional consumption entails but slight additional cost to the company. It seems to me that some of the problems can be solved along that line. There is, of course, some objection to rates of that kind, on account of the complication. Customers often do not understand either the justification for a rate of that kind, or the rate itself. The difficulty is that in making a rate customers do not understand, they can not figure it out. If they can not figure it out, they can not understand it: that is, if you put it on such a basis that they can not understand the meaning of the terms, they can not apply it to the different classes of load. We have been reasonably successful, however, along the line of rates, and it would seem that this method would not only overcome our difficulties in making rate schedules, but would largely remove the legal objections, as it would make

impossible any special rating or discrimination, such as was claimed in the Brooklyn case.

MR. SCOVIL: I wish the gentleman would explain his system on the basis of a one-kilowatt demand, and how he would figure the customer's bill.

THE PRESIDENT: Understanding that the public policy committee has an interesting discussion on rate methods, perhaps it would be well that this discussion does not enter upon that field. The subject under consideration is the legal justification for differential rates. If there is no further discussion we will have the paper on Commercial Development of the Mercury Rectifier, by Mr. Frank Conrad, of Pittsburgh.

Mr. Conrad presented the following paper:

COMMERCIAL DEVELOPMENT OF THE
MERCURY RECTIFIER

Within the last few years considerable attention has been paid to the development of the mercury rectifier for converting alternating into direct current. As it is apparent that this device is to play an important part in the betterment of centralstation economies, a description of some of the latest developments in this line may be of interest at the present time.

The rectifier in its usual form consists of a glass bulb containing two positive electrodes, or anodes, and one negative electrode, or cathode. This bulb is exhausted to a high degree of vacuum and sealed. The two positive electrodes are connected to the terminals of the transformer supplying the alternating current, the connections for the direct-current supply being made between the negative electrode of the bulb and the middle of the transformer winding.

During one-half of the alternating-current cycle, current flows from one of the positive electrodes through the bulb to the negative electrode through the direct-current load to the middle point of the transformer. During the next half of the cycle, the polarity of the transformer terminals is reversed and current flows out of the other positive electrode through the negative to the load. The positives, due to their surface electrode resistance acting as valves, allow the current to flow out, but not in a reverse direction, the resistance of the negative terminal being broken down by the starting operation so that current can flow into that terminal only.

Should the current be interrupted for an instant, the negative resistance will reassert itself and the rectifier will require restarting. Therefore, to bridge over the neutral point of the alternating-voltage wave, it is necessary to give the directcurrent circuit a certain reactance which will store up sufficient energy to maintain the current flow over the zero of the alternating-current voltage.

The negative electrode resistance or valve action of the

electrodes is not an absolutely rigid condition, but is influenced in a number of ways, chiefly in the shape of the containing bulb and the temperature at which it is operated.

To obtain à minimum loss in the bulb, the path between the electrodes should be as short and direct as possible. With a short, direct path, especially at high temperatures, there is a tendency for the electrode resistance on the positives to break down-or, in other words, for the bulb to short-circuit. This may. in a measure, be overcome by operating the bulb at low temperatures, and to obtain this condition, it is necessary either to give the bulb considerable radiating surface or resort to artificial cooling.

The preferable way, especially for high voltage, is to cool the bulb by immersing it in oil contained in a tank having a comparatively large radiating surface. By this means it is possible to use a construction that will give a minimum of loss and still be free from any tendency to short-circuit.

The general design of the bulb and auxiliary apparatus is determined usually by the requirements of the direct-current circuit that is to be supplied, which can in most cases be classed under two heads: namely, the supply of low-voltage approximately constant-potential for charging of isolated storage-battery plants and a high-voltage constant-current circuit for supply of series-arc lamps.

Outfits classed under the first head are usually operated from a low-voltage supply and an auto-transformer is used to transform this voltage to that required by the bulb.

The voltage to be applied across the positive terminals is determined by the direct-current voltage required and the losses in bulb and auxiliary apparatus. As current is drawn from onehalf of the transformer winding only, the direct-current voltage will be one-half the total alternating-current voltage. In addition to this, as the direct-current voltage is the average of the rectified alternating-current waves, while the value of the alternatingcurrent voltage is expressed as effective, it will be reduced by the ratio of the average to the effective, which in a sine wave is 0.9.

The voltage of each half of the auto-transformer in an outfit delivering 110 volts direct current would be 100 plus the loss in the bulb (approximately 15) plus the loss in the choke-coil