When the film is not present there is a plentiful evolution of oxides of iron and titanium, and there is a bright arc. The oxide of chromium can be seen to gather on and finally entirely cover the surface of the pool. This stops the evaporation of titanium and iron, and the light turns to a bluish color and dies down until the chromium film is burned away again. This trouble was met by modifying the mixture in such a way that the oxide of chromium could not separate from the oxides of iron and titanium, thus doing away with the film on the surface and entirely doing away with the dim spells.

In carbon lamps, there was very little done to keep the impurities volatilized from the carbons from depositing on the

globe. This trouble had to be met by the carbon manufacturers, who were prodded up to produce carbons containing less than 0.2 per cent of impurities, but this means was not to be considered in the case of the metallic-arc lamp. The metallic-arc electrodes, being chiefly composed of oxides of iron, titanium and chromium, do not burn away to an invisible gas, as does a carbon stick, but are volatilized bodily, and the vapors instantly condense, on leaving the arc, to a fluffy reddish soot that settles on everything it touches, so that a chimney is a very necessary feature in the lamp. This soot, if it comes in contact with the reflector or globe, will smudge them badly in ten minutes. As was noted before, a current of air flowing down around the electrode served admirably to keep it clean, so it was applied to the reflector and

globe with gratifying results. A thin layer of air is introduced at the top of the reflector and forms a shield through which the soot-laden air can not penetrate, so that the reflector and globe will keep clean for a long time.

As the air currents play such an important part in this lamp, it became necessary to do a large amount of experimental work on the design of an air intake and of a chimney top. The chimney could not be made long enough to cause a very powerful

draft, so the wind was very apt to blow down it; but by persistent effort the openings have been so designed that the wind may blow from any direction (up, down or sidewise), and the only effect is to increase the natural draft in the lamp. Incidentally, the increasing of this draft actually centres the arc and holds it remarkably quiet.

It was found advisable to run the lamp at 4 amperes and 65 to 68 volts at the arc with a cutout set at 85. This low cut

out was made possible by the inverted position of the electrodes and by the peculiar arrangement of the air draft, which prevented any tendency of the arc to flame or to run up the side of the electrode. Without these features, a cutout of 100 to 110 volts would be necessary. As the power-factor at which the lamps operate depends largely on the amount of variation of voltage in the arc, this 85-volt cutout is seen to be very desirable. In actual service, the lamps, including a mercury arc rectifier, run very well at from 65 to 70 per cent power-factor.

In several places above we have described special conditions that must be obtained for getting the best results with these electrodes, and these conditions, of course, must be supplied by the lamp in which they are burned. An example of a lamp well adapted for this service is here shown. A study of its design and construction will show that it is simple and rugged, as no float feed is used; it being necessary only to strike an arc and hold the electrode in a permanent position until, due to change in voltage, the cutout causes the restriking of the arc. The lamp consists essentially of a base and top, connected by a chimney, a set of magnets for striking the arc, a shunt cutout for causing the lamp to feed due to rise in voltage, and a series cutout for disconnecting the striking magnets after the arc has been formed.

The special conditions described with regard to drafts are obtained as seen here by the down-draft tube, which directs the current of air down around the electrode, another current of air over the reflector and circling around the globe forming a means of protection to them, and a special construction of top and case for giving proper draft conditions when under all conditions of wind.

THE PRESIDENT: Discussion on this paper is in order. If there is no discussion I will, on your behalf, thank Mr. Little for the paper, and we will then proceed to the next paper on the programme, entitled Indefinite Candle-Power in Municipal Contracts, by Mr. E. Leavenworth Elliott, of New York.

Mr. Elliott presented the following paper:

'INDEFINITE CANDLE-POWER IN MUNICIPAL

CONTRACTS

It is doubtful if there is any legal instrument that is a more fertile source of litigation than a contract. Even the most studied and painstaking work of minds highly trained in the technology of law is never proof against misunderstanding, misconstruction, and loopholes for evasion. The inherent weakness of a contract is a direct result of the difficulty of two different minds making themselves mutually understood. The phrasing of a contract is an attempt to express a "meeting of minds," and in the final test its true meaning can be determined only by a careful consideration of all the circumstances that tend to show the actual basis upon which the two minds met. While it is absolutely impossible to guard against differences of opinion as to the meaning of a contract, the very existence of such a difficulty furnishes an urgent reason for the most explicit statements and rigorous exactitude in the use of terms. This applies particularly to terms that are used with a special or technical meaning.

There are in existence what are popularly supposed to be lighting contracts, involving the payment of many millions of dollars annually, which if carefully scrutinized will be found not to be lighting contracts at all in the strict sense of the term, but contracts for the sale of electric current in connection with the use of certain apparatus for generating light by means of the current furnished. It may be broadly stated that light has never been sold as a commodity; that is, the payments made have never been based upon a measurement of the actual quantity of light furnished. This is a result of the simple fact that, up to the present time, there has been no means of measuring light with sufficient accuracy to make such a basis of payment possible. Light, being a form of energy, can be measured as a commodity only by taking it in connection with the measurement of time; and an instrument for integrating luminous flux with time has, so far as we are aware, never been even attempted, much less perfected. All light sources fluctuate in the quantity of light given out, and this fluctuation has been one of the chief obstacles

in reducing photometry to an exact science. Of the various commercial light-sources, the electric arc is notoriously the most variable; hence the difficulties in expressing its light-giving power by any constant quantity. All attempts that have been made involve a method of averages but entirely neglect the element of time; and if an are varies between, say, 100 and 400 spherical candlepower, it by no means follows that the average flux of light throughout the period is the arithmetical mean of these quantities. It may give out light at the maximum power for three-fourths of the time, and at the minimum for only one-fourth, or vice versa, or in any other ratio.

In view of these inherent obstacles in the measurement of light, and especially from the electric arc, it is not surprising that many so-called lighting contracts exist in which the amount of light to be furnished is a wholly indeterminate quantity; and it is somewhat surprising that such contracts have not given rise to more litigation than they have. When the arc lamp first made its appearance it wholly baffled the attempts of photometrists, with the means then at their disposal, in their attempts to give it a candle-power rating that should give even an approximate idea of its actual light-giving power. It is related that when one of the first commercial arc lamps was completed it was examined by a college professor, who was asked to give an estimate of its candle-power. He replied in an offhand manner that he thought it would run about 2000; and an arc lamp of that particular type has been called a 2000-cp lamp even unto this day. Although it was soon learned that such a rating was enormously exaggerated, the fiction, having once come into use, persisted to such an extent as to find its way into many lighting contracts. In order to prevent the term being accepted at its face value by the uninitiated, a committee of this association took the matter in hand and gave a specification for lamps of so-called "nominal," or "standard 2000 candle-power." While this committee represented only one party to lighting contracts, the definition that it formulated has been generally respected as an avenue of escape from the difficulties of the situation. Where the lamp thus described was maintained in contracts calling for a 2000-cp arc light, the acceptance of the definition served to prevent legal controversies; but in the progress of the art of lighting the particular form of lamps that was officially rated as 2000 candle-power has become