at the end of 96 hours there is a decrease, so that in all probability a complete series of tests would show that the deterioration of this mantle was practically uniform, the same as is shown by No. 3.

No. 4 exhibits the same characteristics as No. 1, namely, a very rapid falling off in initial candle power and a more gradual decrease afterward. The rapid decrease in No. 4, however, is greater than in No. 1.

Plate 2 shows the characteristics of these four mantles with reference to their efficiency or the candle power per cubic foot of gas at different periods of their lives. This plate shows, as do the other tests, that mantle No. 1 is of much higher efficiency than the others throughout its entire life, and that No. 4, starting off with a high initial candle power, falls below the others after less than 100 hours burning. No. 2, although starting lower in candle power than any of the other mantles, is superior, with the exception of No. 1, throughout most of its life. Plate 2 gives perhaps as good an idea as is possible to obtain, graphically, of the relative values of these mantles, and shows that their values are, roughly speaking, in accordance with their costs. It is to be noted that there is, on the whole, comparatively little difference between No. 2 and No. 3, as far as the light-giving quality of the mantles is concerned. A breakage test, however, would show that No. 2 is superior to either No. 3 or No. 4, and No. 3 is better than No. 4. We thus see that the value of a mantle depends, not only on its candle-power hour performance, but also on its ability to withstand the usage which it is ordinarily subjected to.

It will be noticed that No. I deteriorated in candle power per cubic feet of gas in the 555 hours of test about 24% and Nos. 2 and 3 about the same, showing that these three mantles deteriorated about equally. No. 4, however, representing the cheapest mantle tested, showed a deterioration of no less than 46%. The net low efficiency throughout the life of the No. 4 mantle is largely explained by the fact that the skirts of these mantles at the cap wore off at the bottom until no longer overlapping the cap. In some part, the wearing away of the skirt of the mantle may be explained by the fact that after these mantles were burned off, in most cases they were no

longer cylindrical at the bottom, the distortion causing them to rub against the cap. One thing to be carefully noted from these tests is that sometimes the cheapest mantles on the market have a high initial candle power, but that they very rapidly fall off in value. This class of mantles is largely handled by peddlers, and even by some of the gas companies in their endeavors to increase the sale of gas. The practice of putting out such mantles as these is decidedly objectionable, inasmuch as the customer, starting off with a mantle of high initial candle power, is apt to blame the gas company for the very rapid and marked decrease in light, which leads to dissatisfaction and opens up an opportunity for the electric company.

As before stated, the results of these tests are unsatisfactory. The question as to the value of mantles is a very broad and comprehensive one, and is not or should not be determined by any one individual. The question being, in the opinion of the writer, of such vital interest to the gas industry of this country, he would suggest that it is fitting for the new American Gas Institute, representing, as it does, almost the entire gas industry of the country, to appoint a committee which would thoroughly investigate the subject, the proper funds for the same to be supplied by the Institute. Complete tests would be too expensive for one individual or company to carry on, over 100,000 cubic feet of gas being used in the tests given in this paper, while if shared equally by those who are interested, namely, by the members of the Institute, the pro rata cost would be small. Further, if such tests were conducted individually, they would lack the authority which would be given them by being conducted by the Institute.

A similar committee was appointed some years ago by the National Electric Light Association to investigate different forms of illuminants, especially are lights for street lighting, and the results of those tests were of great benefit to the members of the Association and the industry as a whole, and remain today the standard in this country. Would it not be possible for the Gas Institute to employ competent engineers to carry on such tests, so that the whole gas industry would reap the benefits. At the same time this would remove all questions of commercial bias. It might even be possible to establish a

permanent laboratory under the auspices of the Gas Institute, where mantles selected from any given purchase could be sent and tested, such a laboratory to fulfill in the gas field the same place as the Electrical Testing Laboratories fulfill today in the electric field. Such a committee could undertake to standardize a set of tests or specifications which mantles of a given price or quality should undergo. The writer would tentatively suggest that such a committee consider the following:

I. The initial candle power on a standard gas, at a given pressure and with a given consumption, should be not less than a stated amount.

2. The life of a mantle should be a stated number of hours before it falls to a given percentage of its initial candle power.

3. A mantle should undergo a jar test to prove its mechanical strength. A standard jar test could be easily arranged, so that all mantles would be subject equally to such vibrations.

4. The method of support for mantles should be stated. Tests could probably be conducted to determine the best method of support.

5. The weaves of mantles could be specified. Complete tests would undoubtedly show which weaves are best adapted for different conditions.

6. The amount of shrinkage during the useful life of a mantle should be distinctly specified, in order to prevent binding at the skirt, etc., with consequent breakage.

7. The question of shape both before and after burning off should be carefully considered, as many mantles fail from such causes.

8. The question of color should be carefully considered, and specifications should state the color required, as different colored light is required for different purposes.

If rigid specifications with a penalty clause were adopted by gas companies, and the Institute or some standard laboratory could conduct such tests, it would be possible to select, say one mantle at random out of each thousand, with the understanding that such a mantle represents the thousand. If such a mantle falls below requirements, the entire lot would be penalized.

If the large gas companies could buy mantles on some such