world's most valued reference standards of length and weightthe international meter and kilogram-together with certain "témoins" or ancillary duplicates. The building is mainly devoted to comparisons between these fundamental standards and the corresponding replica of other nations, through the medium of working copies. It has no upper stories and is surrounded by the forest park. Fortunately for the even tenor of its duties, it ordinarily escapes attention from the ubiquitous tourist, despite its proximity to the much visited porcelain museum. Neither Baedeker nor "Guide Bleu" refers to it, and it finds no place on the list of objects to attract the interest of the visitor; yet it may properly be described as a latter-day temple for the guardianship of the "Lares and Penates" in the world of weights and measures. The expense of its upkeep is shared, in definitely prescribed proportions, among the twenty-eight foremost countries of the world.

The need for an international clearing house and depositary for the world's standard meter becomes evident from a consideration of industrial needs alone, as well as from a cursory glance at the history of the subject.

If a steel bar, say one yard in length, is ordered, without any further specifications, from a smithy, we might reasonably expect the bar to be delivered true to length within one per cent., more or less; that is to a precision of one part in 100, or 102. This may be described as a precision of the second order. It does not mean that the smithy could not furnish a higher degree of precision in length, if the need existed; but merely that in the ordinary course of business, a yard bar at the smithy might well be interpreted to include bars longer or shorter than a yard by as much as one per cent. Moreover, the measurement of the bar, by the act of laying a yard-stick alongside it, would be an operation lasting only a few moments. Nevertheless, if the tolerance of something more than one per cent. were clearly admitted, the moral certainty of the measure within those limits would be very great. Any intelligent man receiving the bar from the smithy, and laying a creditable yard-stick beside it, could see at a glance that the two were nearly alike in length; so that unless some question as to the amount of tolerance were raised, he would feel a high degree of assurance that—in the common use of language the bar was a yard long.

The rough bar from the smithy might next be sent to a machine shop with an order that it be trimmed, smoothed and cut to a finished length of say 35 inches; but without any specifications as to tolerance in precise length. In the operations of smoothing and finishing, the mechanic entrusted with the order, would probably

spend a little time in adjusting its length to "35 inches." One per cent. he would regard as unworkmanlike, and he would probably aim at a precision say of one per mil; i. e., one part in 103, or of the third order. If he were put upon his mettle, he might be willing to attempt a precision of the fourth order, or 1 in 10*; but the extra time involved in such an effort might not be justifiable. Even to reach the third order, he would have to make the length correct within 0.035 inch, and much more time would be needed in the measurement than was spent at the smithy in attaining the second order. In the first place, an ordinary rough yardstick would no longer suffice. A graduated steel tape or straight edge would be necessary. Moreover, the operation of juxtaposition and reading off the length would take much longer than before. Finally, in spite of this increased mental and physical care in the task of measurement, the degree of moral certitude as to the reliability of the result will probably not have increased proportionately. In repeating the measurement, the mechanic will begin to arrive at slightly different results and the effect of the discrepancies disconcerts the judgment. Of course, unless the mechanic made a gross error or mistake in reading the tape, he would feel convinced that the finished bar was much nearer to "35 inches" than one per cent.; but upon the new plane of third- order precision engaging his attention, he might feel less assurance of success than his predecessor, who measured only to the second order of precision at the smithy.

If now the finished 35-inch bar were sent to a superior workshop, to be assembled perhaps in some fine piece of mechanism, after being adjusted to say 34.8 inches, at a room temperature of 20° Centigrade, with a tolerance of 0.004 inch, this would entail refinishing the ends of the bar and measuring its new length to one part in 10, or to fourth order precision. For the purpose of making this measurement, a special gauge might have to be prepared. If the bar in the assembled mechanism had to be made interchangeable with similar bars coming from other machine shops, the gauge might have to be adjusted to a precision of the fifth order in length, at a special workshop for the construction of precise gauges. In order to reach fifth order precision in the gauge at the special workshop, a standard measuring rule of yet higher precision would be needed there. It would not be unreasonable to require a precision of the 5th order in that special workshop standard. That standard would probably be compared and calibrated against a still finer standard at the Bureau of Standards in Washington, where a precision of the 6.5th order, or one per 3,000,000 might be readily obtainable. Finally, to keep the stand

ards of the different national laboratories of the world in mutual agreement, an international standard is maintained at Sèvres, where the precision attainable is of the seventh order.

It is clear that every measure of length, either in the world of business or in the world of science, has its order of precision, over the entire range from the first to the seventh, depending upon its construction and purpose. No one expects a higher order of precision than the particular purpose of the measure in question demands; because the effort and expense involved in securing an extra order of precision is relatively great. There are certain lines of industry, notably in gauge making for fine tools, where fifth order precision is necessary. Not many decades ago, this was the highest order scientifically attainable in measures of length. The advance from the fifth to sixth order demanded an immense amount of scientific and industrial effort. Progress had to be made in the mathematics of accidental errors, in metallurgical chemistry to provide improved materials, in physics to learn the laws of length variation in material standards, in tools, to fashion the improved parts, in workmanship and experience to handle the new tools. At the present date, we can look for the seventh order precision in the comparison of the various national meters with the international meter at Sèvres; but although this suffices for practically all industrial needs, it is inadequate for certain scientific requirements. Certain problems of the Einstein theory, for instance, might find solution, if the eighth or ninth order of precision in the measurement of length were attainable.

If material civilization advances, we may hope to secure one higher order of length precision at Sèvres in the course of another century. This would probably add one order to national scientific length measurement all over the globe. Apart from questions of moral or spiritual development, an estimate of the world's civilization might be furnished, based upon the order of precision realizable in the certificates of meter-bar comparisons furnished by the bureau at Sèvres.

The permanence and inviolability of the international meter are clearly of importance to all nations. The control of the Sèvres bureau has been vested, since 1875, in an international body composed of delegates from the twenty-eight leading national governments that are parties to the bureau's maintenance. The international meter and kilogram are deposited at the bureau in such a manner that seven successive keys have to be used in order to reach them. Three of these keys are in regular service for the doors of the building; but the four others belong to the special vault in which the standards are preserved, and are placed in the

custody of as many different officers of the international committee. Some of these officers live abroad; so that the keys are usually kept far apart. It has thus only been possible to open the vault at the regular six-year meetings, when the assembling officers produce their respective keys. The inconvenience of so inflexible a modus operandi manifested itself during the world war. At the outbreak of the war, one of the officers, entrusted with a key, resided in Germany. That key, being in an enemy country, was inaccessible. If Sèvres had been subjected to bombardment, it would have been necessary to remove the standard meter and kilogram to a place of safety, and this would have involved breaking into the vault. In order to provide against such a contingency in future, a new set of regulations has been brought into effect; whereby a duplicate of each vault key is deposited with the Institut de France, and so that, under special emergency, the vault can be opened by its authority.

One of these six-year meetings of the International Conference took place recently in Paris under the presidence of M. Emile Picard, the permanent secretary of the French Academy of Sciences. On October 6th, 1921, the delegates met at Sèvres and at a specified hour formed themselves into a visiting and attesting committee, under the leadership of M. Ch. Ed. Guillaume, the Director of the Bureau.

The committee members line up across the courtyard in column by twos. At the signal, the procession enters the main and east door of the Bureau, and passes along a corridor, in the half light, to a descending stone stairway. Incandescent lamps light up, and we descend to a basement floor. Again, down another stone stairway, to a sub-basement. At this level, the temperature changes but little all the year round. We now face the first of the three steel vault doors in the east wall. It opens to the corresponding official key. Behind it are two other steel doors, which are successively unlocked, revealing the vault beyond. This is about 4 meters long in an easterly direction, 3 meters wide and three high. An electric incandescent lamp, that has been idle for eight years, is turned on and we can see the interior clearly. The walls and ceiling are lined with white enamelled brick. Opposite to the entrance against the eastern wall is a table supporting a steel safe. There is nothing else in the vault except an auxiliary table against the north wall. The director produces the last of the seven keys and unlocks the safe. Its doors swing open and disclose two shelves. On the upper shelf are three meter-bar cases, a minimum-maximum thermometer and a hygrometer. On the lower shelf is a row of five glass double bell-jars. Inside of each is a shin

ing cylindrical standard kilogram. The director calls for a reading of the instruments, which have been shut up in the safe since 1913. The min-max, thermometer register 10.6°-13.2° Centigrade, or a total range of only 2.6 degrees during those eight years. The hygrometer shows 88 per cent. humidity. The director calls attention to the international meter and kilogram, with their control duplicates. He carefully lifts out of the safe the central case, containing the primary standard or prototype meter, and lays it on the auxiliary table. He opens the case and reveals the brightly gleaming meter bar within. Like all the other standard bars, its section is of a special X shape, so designed as to offer the maximum stiffness, or resistance to sagging in the middle, when the bar is supported at its two ends. Of course, a stout platinum-iridium bar, a little more than forty inches long, and built with any reasonable shape of cross-section, would sag very little at the center when the bar is supported at its ends. Nevertheless an extremely small sag would be apt to alter, in perceptible degree, the apparent length of the bar, as measured on its surface. By giving this X shape to the section, and cutting a flat strip of surface at the middle of the groove in the X, the meter length is marked off along this flat strip, where the change in length due to any possible central sag becomes quite negligible. The meter bar is not graduated, or marked off into equal divisions like an ordinary rule. There is merely a fine scratch cut with a diamond point across the bar near each end. The international meter is defined as the distance between these two fine line scratches, when the bar is at the temperature of melting ice.

No hand touches the bar, which lies face down in its case. It is the final reference standard, and does not need to be used except as a final arbiter, in case differences should arise among the working standards. Its duty is merely to remain steadfast-to preserve its dimensions unchanged.

Only a few of the witnesses can occupy the vault at one time, and the air in it becomes oppressive. As soon as those within have recognized the contents of the safe, they leave the vault and make room for others.

In about half an hour, the standards are replaced on their shelves, the safe is shut and locked, the light is turned off and the vault doors closed. The world's standards of length and weight thus resume their wonted repose, until the next awakening by a visiting committee, probably six years hence. This visiting committee, however, returns to daylight and reforms in the courtyard, where it is photographed by a moving-picture machine, to record the passage of its members.