ber wall on O-ring seals. The clear apertures were reduced to areas about 13 mm high by 8 mm wide by masks.

The vacuum connections to the vacuum chamber were made through a special flange and O-ring seal near the midheight of the chamber. A copper tube connected to the flange led to a 19-mm valve supported by the chamber. A detachable connection from the valve permitted coupling the system to a mechanical pump by a flexible hose. The system was designed to present a minimum throat diameter of 19 mm. The vacuum pump was a two-stage mechanical pump with a free air capacity of 140 liters per min.

The vacuum chamber was equipped with a thermocouple-type vacuum gage inserted in the chamber wall at a point about 24 cm below the connection to the vacuum pump. Once the vacuum chamber had been outgassed it was possible to attain a pressure of about 0.13 N/m2 (1 μm of mercury). The vacuum pump was operated continuously except during the length or free-fall measurements. Generally the leakage rate for the system ranged from about 0.4 to 0.6 N/m2/hr at this pressure.

A mercury-in-glass thermometer was imbedded in the thermal insulation of the vacuum chamber near the vacuum gage. The thermometer bulb was kept in thermal contact with the brass wall of the chamber by a pad of compacted aluminum foil. The thermometer was covered by the cork insulation except for a slit exposing a portion of the engraved scale. The ther

mometer scale was graduated to 0.2 deg Celsius and read by estimation to 0.02 deg. This thermometer was identical to the thermometers used to measure the temperature of the length standard.

2.4. Mechanical Design

The frame of the apparatus for the determination of the acceleration due to gravity consisted of a base plate of steel 120 cm square and 3.8 cm thick, grouted to the floor of the building and held down by four 2.5-cm anchor bolts. At each corner of the plate there was a hollow square-section steel column 3.7 m high. Bridging the tops of the columns was a plate of steel 120 cm on each side and 5 cm thick, cut out along the edges to reduce the weight. Two guide rods for the carriage were supported in the frame gripped in rings bolted to the base and upper plates. The internal diameter of each ring was 7.5 cm; each ring had four centering screws. The guide rods were 6.35 cm in diam with a tolerance of 0.1 mm. They were selected to be straight to within 0.08 mm per m of length. The spacing between the guide rods was nominally 40 cm.

The carriage of the apparatus was about 117 cm long and was formed of two aluminum alloy plates separated by vertical tubes of aluminum alloy as shown in figure 3. The two guide rods passed through holes in the carriage plates and through the interior of the two aluminum alloy tubes.

At both ends of one tube two pairs of ball bearings were mounted so as to center the steel guide rod within the aluminum alloy tube. A single pair of ball bearings at the upper end of the opposite tube constrained the carriage against rotation about the opposite guide rod. The arrangement of the ball bearing guides is shown in figure 5. The five pairs of ball bearings were adjusted against the guide rod with a slight negative clearance to eliminate any possible

lateral freedom of motion. The motion of the carriage was found to be extremely linear and yet with only about 100 g friction in the direction of free fall.

Since it was necessary to counterbalance the weight of the carriage and vacuum chamber about 22 kg during length measurements, a pair of ball bearing pulley wheels were attached to the upper main frame plate near the vertical extension of the ends of the carriage frame. Weight hangers were connected to two spring steel ribbons 12.7 mm by 0.18 mm which passed over the pulleys and terminated in fittings that could be coupled to the frame of the carriage. A view of the assembled frame, carriage and vacuum chamber is shown in figure 6. The wood platform for the observer was supported from the building floor and did not touch the gravity apparatus.

The main guide rods were alined parallel to the direction of gravity by the following procedure. Near each guide rod a fine wire was suspended from the upper plate of the frame. A mass of sufficient weight to stress the wire moderately was attached to the lower end of the wire. The lower end of the wire and the mass were immersed in a pot of oil to damp out oscillations. A microscope with reticule graduated to 0.01 mm was mounted on the counterbalanced carriage and focussed on the wire, first in a plane perpendicular to the face of the carriage and later in a plane parallel to the face. Microscope readings were taken on the wire at the upper and lower limits of travel of the carriage and the position of the guide rod corrected by adjust

ing the centering screws in the retaining rings unti the motion of the carriage was parallel to the plumb line to within 0.02 mm in one m.

The release latch and initial accelerating spring were mounted on a subassembly securely attache to the upper frame plate of the gravity apparatus. Th arrangement of the subassembly is shown in figure 7 The latch was designed in such a manner that th carriage received no lateral torque or impulse whe the latch was released The latch parts were groun and finally lapped in place to provide a smooth, clea separation at release.

The portion of the latch attached to the carriag was coupled to it through a bridge frame anchore at the upper ends of the two aluminum alloy tubes th formed the vertical members of the carriage (fig. 9 This method of mounting insured that the upper pla of the carriage would not be subjected to bendi due to suspension at its center. Consequently the would be no sudden release of stored energy to transmitted directly to the vacuum chamber.

When the carriage was suspended from the lat it was also in contact with the plunger of the init accelerating spring. The accelerating spring assem consisted of a precompressed helical spring contain in an adjustable barrel with provision to adjust b the force exerted by the spring and the vertical tra over which the plunger remained in contact with carriage. These parameters controlled the amount separation between the silica tube and the vacu chamber support cone. As adjusted in practice, spring exerted a force equal to approximately percent of the weight of the carriage and follow the carriage for about 8 mm in its downward tra after release of the latch. The spring assembly appe as the knurled and threaded barrel located just ab the latch in figure 7.

The range of the acceleration spring adjustment such that it was possible to adjust from a condi where the silica tube separated only momenta

from the supporting cone to a condition where the last aperture on the silica tube was no longer in line with the vacuum chamber optical window at the time it passed the sensing station. The actual operating condition was chosen at an intermediate value where the observed transit times were independent of the accelerating spring adjustment.

The vacuum chamber was attached to the carriage of the apparatus by centering screws so arranged that the upper and lower flanges of the chamber each rested against two adjustable screw stops on the longitudinal and transverse axes of the carriage. At each end a locking screw acting along a line at 45° to x the screw stops held the flange in firm contact. This arrangement permitted the entire vacuum chamber to be removed from the carriage and then replaced e without disturbing the alinement.

In the process of alining the vacuum chamber in the carriage, rotational orientation was established by observing the superposition of the images of a point light source reflected from the upper optical window of the chamber and from a glass flat in contact with the upper cross plate of the carriage. The optical axis of the sensing station was alined to be perpendicular to the same reference plane.

The apparatus was provided with a pneumatic receiving cylinder on the lower plate of the frame and a plunger attached to the lower crossplate of the carriage to bring the carriage to a stop at the end of a its fall. The plunger was machined to fit the cylinder with a radial clearance of about 0.5 mm. A lubricated O-ring in a groove in the plunger completed the air seal between cylinder and plunger. The receiving cylinder was 56 cm long and 10 cm in diameter. A 5-m length of high-pressure type rubber tubing was connected to the cylinder through a port at the lower end. The other end of the tubing was connected to a closed-end section of larger diameter tubing. This combination provided damping to reduce the tendency of the carriage to rebound after being brought to a stop.

2.5. Optical System

The optical system of the position sensing aparatus is shown in figure 8 with the photomultiplier cell housing slightly retracted. This assembly was mounted on a steel bracket attached to a concrete pier directly behind the gravity apparatus (see figs. 3 and 6). The lens, a 100-mm microscope objective lens, focussed an image of the aperture plate edge on a slit mounted in the focal plane of a microscope eyepiece in the viewing system housing. This arrangement permitted an observer to focus the instrument by eye and to aline the slit with respect to the image. Once the alinement was completed, the optical system was terminated by the photomultiplier cell mounted in the sliding housing and making a light-tight seal to the viewing housing. The slit in the viewing system was approximately 4.5 mm long and 0.075 mm wide. The width of the slit was not critical and could have been much narrower since the output of the photo multiplier reached a maximum value in a 10-μm travel of the silica tube. As the magnification of the viewing

FIGURE 8. Optical system with photomultiplier assembly.

lens was approximately unity, a slit 10 to 12 μm wide would have been adequate. The wider slit was chosen in preference, however, since during the preliminary adjustment by eye, a narrow slit produces diffraction patterns making alinement and focussing of the images difficult.

The photomultiplier tube used was a 10-stage, head-on tube having a maximum response in the spectral region of 400 to 500 nm. The tube was operated at a cathode potential of 900 V. The photomultiplier amplifier. The coupling amplifier provided a lowtube anode was connected directly into a coupling capacitance, 51,000 load on the photomultiplier anode. The anode potential was 6 V. The output of the coupling amplifier was connected to a 90- terminated coaxial cable leading to the timing system.

The light source was an incandescent filament, galvanometer lamp operated at 6 V, direct current. The lamp had a straight, helical coil filament. The lamp was housed in a case having a lens on a focussing mount. The lens projected a roughly parallel beam of light which, with a vacuum chamber window in line with the axis, passed through the chamber window, through the transparent portion of the fused silica. tube and provided a uniform illumination to the aperture edge plates mounted on the opposite side of the silica tube. The light source was mounted on a special telescope stand to facilitate levelling and positioning.

2.6. Electronic System

The power supplies employed for the apparatus included a low voltage direct current power supply for the light source and a high voltage direct current power supply for the photomultiplier tube. The signals from the photomultiplier tube were coupled by a coaxial cable into the input amplifier of a fastrise cathode ray oscilloscope. The triggering circuit of the oscilloscope was adjusted to start the sweep circuit at some potential level of the linearly rising (negative) signal from the photomultiplier tube. The

timers were then triggered from the sawtooth sweep circuit of the oscilloscope. This arrangement offered a high degree of flexibility in adjustment plus very good isolation between the timers and the photomultiplier tube.

The timers were especially designed for the apparatus and sequence of signals employed. A master unit contained all the gating circuits required, a 1-MHz thermostatted quartz crystal oscillator coupled to a 10 frequency multiplier, the shaper, and a sevenplace decimal counting unit. A second slave unit carried another seven-place decimal counting unit. Both counting units had a least count of one-tenth of a microsecond.

The timers were so coupled that on receipt of a signal from the oscilloscope the gates to both the master unit and the slave unit were opened, starting both decimal counting units. Upon receipt of a second signal, the gate to the decimal counting unit in the master timer was closed, leaving the gate to the slave unit open. The receipt of a third signal closed the gate controlling the slave unit.

The timers were provided with an optional circuit such that an interval displayed on the indicator panel could be retained while the gates were reset for a new timing sequence. Subsequent intervals were then added to the intervals already displayed. This circuit was not employed in the measurements on the falling silica tube but was utilized in assessing the errors of the gating circuits as described later.

3. Reference Standards

3.1. Length

The standard of length for the determination was an invar, 1-m line standard having an H-shaped crosssection. The scale was fabricated and ruled by the Société Genevoise d'Instruments de Physique of Geneva, Switzerland, in 1959. The scale was ruled in 1-mm intervals with line widths of about 5 μm. The coefficient of expansion of the invar was determined as 0.00000123 per degree Celsius.

The lengths of 0-300 mm and 0-1000 mm intervals on the scale have been compared to the standards. of the United States periodically since the scale was received in February 1960. The results are shown in table 1. It was considered that the scale was very stable for an invar standard and that the results indicated no progressive drift in the lengths. Accordingly, the means of all of the reported corrections have been applied to the measurements of the fused silica tubes in the gravity determinations.

3.2. Time

The standard of time for the determination was based on the National Bureau of Standards frequency standards, since only time intervals were measured. The crystal controlled oscillator of the master timer was compared to a 1,000-Hz signal derived from the Bureau's quartz oscillator, which, in turn was com

pared on a daily basis to the WWV frequency standar The comparison of the master timer with t 1,000-Hz reference signal was accomplished by sta ing the timer with a single sweep of the oscillosco triggered from the reference signal. After an inter of approximately 100 seconds the oscilloscope swe was reset and again triggered by the 1,000-Hz sign Even though the interval displayed was not exac 100 seconds, it was some integral multiple of 1,000 Since the difference between the crystal oscillator the timer and the frequency standard never exceed 50 μs in 100 s, there was no ambiguity in determin the relative error of the timer frequency base.

The crystal oscillator was checked prior to e free-fall determination and again after it. Althou as indicated above, the oscillator was always ma tained within 5 parts in 108, the observed correcti were applied to all time interval measurements.

4. Techniques of Measurement

4.1. Length

Measurements of the two lengths, L1 and L2, w carried out using the same apparatus employed the free-fall determinations. The weight of the carr was balanced by means of the steel tapes running the frame pulleys with appropriate weights on other ends of the tapes. The invar length stand was attached in holders beside the vacuum chan in such a position that the rulings on the specu surface could be viewed by a reading micros attached to the support table for the position sen apparatus. A fine motion drive was attached to a g rod and served to advance the carriage by the motions required.

Basically the procedure consisted of moving carriage to a position where the lower apertur the silica rod was in a position with respect to sensing station such that the output signal from photomultiplier cell was at the same potential as which would trigger the oscilloscope-timer com tion. At this point a reading was taken on the ruling of the invar scale with a filar micromete the reading microscope. The carriage was moved on to bring the second aperture into line the sensing station and a microscope reading t

on the 300-mm graduation of the invar scale. The process was repeated again for the upper aperture and the 1,000-mm graduation.

It will be seen that this process is one of measuring how far the silica tube must be moved to produce the triggering signals rather than of measuring its length. This method of measurement has the advantage of including the effects of vacuum on the silica scale as well as any possible displacement of the images by the optical windows of the vacuum chamber. More important, since there are complex diffraction effects at a shadow edge, the measurement is related directly to the conditions present in the experiment and measures directly the quantity sought.

The reading microscope used for the observations on the invar scale was mounted on the base plate for the optical system. Provision was made for the microscope base to be attached on either side of the position sensing system. The microscope was mounted on a special base with a rack and pinion focussing control and a fine adjustment screw. The microscope tube length was 215 mm with a 22.5-mm objective and a 12.5x-filar micrometer eyepiece. The filar micrometer eyepiece was equipped with double cross-hairs spaced the equivalent of 12 μm in the object plane. The micrometer drum was graduated into 100 divisions per revolution. The reading microscope was calibrated on an auxiliary scale and had a factor of 1.112 μm per division. During observations the micrometer drum was read to one tenth of a division. In no case was it necessary to measure differences exceeding 50 divisions with the filar micrometer eyepiece.

During length measurements it was necessary to locate the carriage in a position where the output from the phototube amplifier was at the potential required to trigger the timing system. A direct current voltmeter with a trimmer was connected in parallel to the output of the amplifier and adjusted so that it indicated full scale at the maximum output of the amplifier. The triggering potential was then set to be at the half scale reading. The indication was such that the small graduations on the voltmeter corresponded to movements of about 0.2 μm at the carriage. The position could be set to a small fraction of a voltmeter division with the aid of the fine-motion drive. The meter was slow enough in response to mask the effects of photomultiplier tube noise and building vibration.

The fine-motion drive which made possible the adjustment of the carriage to the final position for reading the length standard consisted of a hinged brass clamp which could be opened, slipped around a steel guide rod and locked in place. A bronze spring provided sufficient frictional drag on the smooth rod to hold the clamp against the 100- or 200-g unbalanced weight of the carriage, yet permitted the clamp to be slid by hand as a coarse adjustment. Attached to the clamp was a flat wedge, driven by a fine thread screw, actuating a small push rod to move the carriage.

4.2. Temperature

The invar length standard was equipped with two mercury-in-glass thermometers, one mounted near

the center of the 0 to 30-cm length and the other near the upper end of the scale. The thermometers were graduated to 0.2 deg Celsius and had errors less than 0.02 °C at the ambient temperature of the room (22 °C). The thermometers were mounted against the back surface of the web of the H-section of the length standard; the bulbs were packed in aluminum foil compressed against the surface of the invar and covered by 6 mm of cork sheet up to the graduated portions of the stems. The exposed portions of the cork pads were covered with a layer of aluminum foil.

A similar thermometer was mounted in contact with the lower section of the vacuum chamber. The chamber was covered by a 6-mm layer of cork sheet topped by a double layer of aluminum foil. While it was recognized that the temperature of the silica tube was not necessarily the same as the temperature of the chamber wall, it was assumed that the tube temperature maintained a reasonably constant relationship to it, once thermal equilibrium was established. The temperature of the vacuum chamber walls during length measurements averaged about 0.2 °C less than that during the free fall measurements. This difference probably was caused by the larger number of pieces of electronic equipment operating during the free fall measurements.

4.3. Compensation for Guide Curvature

The two guide rods, previously described, on which the carriage ran were of a special case-hardened steel, centerless ground and selected for straightness. The diameters were 63.47±0.020 mm and the rods were straight within 80 μm per meter. These tolerances were quite adequate for operation in the free-fall phase of the experiment but the slight curvature of the rods was sufficient to make an appreciable error in the length measurement. The error is proportionate to the product of the angle of rotation of the carriage and the distance between the axis of the invar scale and the silica tube.

If the position of the silica tube and the invar scale are interchanged with respect to the direction of rotation, the algebraic sign of the error is reversed. Thus the mean of two measurements with the scale interchanged between them should be free of the error due to curvation of the guide rod, provided that the rotation of the carriage is the same in both sets of measurements.

The carriage of the gravity apparatus was provided with duplicate sets of holders for the invar scale, one set on each side of the vacuum chamber equidistant from the centerline of the silica tube. A length measurement consisted of two sets of measurements, one for each location of the invar scale. The final computed lengths from each set were averaged to obtain a working value. Ordinarily the first set of measurements was made at the start of a day. Upon completion of the first set, the invar scale was transferred to the opposite set of holders: about five hours later the second set of measurements was made. The order in which the two positions were occupied was reversed on alternate days of length measurement.