``Bicentenary of the Cavendish Experiment" Conference

Riley Newman, University of California Irvine
rdnewman@uci.edu

The measurement of G has become a popular occupation following the announcement a few years ago of widely discrepant new G values -- most notably a value reported by the German PTB (NIST-equivalent) lab which is more than half a percent away from the accepted "CODATA" value (see volume 4 of this Newsletter). Thus the celebration of the 200th anniversary of Cavendish's G measurement with a conference in London last November, sponsored by IOP and organized by Terry Quinn of the BIPM and Clive Speake, drew a lively crowd of G measurers to Cavendish's turf. Twelve groups actively pursuing G measurements were represented; six of these announced new or updated G values:

Lab G x 1011 (ppm) $ G-G_{CODATA}\ (\sigma) $ $ G - G_{PTB} \ (\sigma) $
New Zealand MSL 6.6742(6) 90 1.5 -51
Zurich 6.6749(14) 210 1.4 -27
Wuppertal 6.6735(9)(13) 240 0.5 -25
JILA 6.6873(94) 1400 1.6 -3
BIPM 6.683(11) 1650 0.9 -3
Karagioz (Russia) 6.6729(5) 75 0.3 -57
Luther/Towler 1982 6.6726(5) 64 - -58
PTB 1995 6.71540(56) 83 42 -

Here the comparisons with CODATA and PTB in standard deviations reflect uncertainties in both numbers compared. The last two lines give the 1982 result of Luther and Towler (on which the CODATA value is based, after doubling the assigned uncertainty), and the puzzling 1995 PTB result.

All six new results are higher than but roughly consistent with the CODATA value. The PTB value remains a mystery, not to be lightly dismissed -- the CODATA committee will have difficult decisions to make in its next round of assessments! No lab yet feels it has surpassed the 64 ppm accuracy that Luther and Towler assigned to their 1982 measurement, although a number of groups target accuracy of 10 ppm or better.

The twelve approaches to G measurement are remarkable in their variety - no two are very similar in technique. Heedful of Kuroda's caution about the perils of anelasticity, all but two of the experiments either avoid the use of a torsion fiber or use a fiber in a mode such that its internal strains are negligible. The New Zealand MSL lab compares the torque on a torsion pendulum with that due to an electrostatic force which is in turn calibrated in terms of the angular acceleration it produces on the pendulum in a separate experiment. The Zurich group uses a beam balance to weigh kilogram masses in the presence of mercury filled steel tanks - this group anticipates greatly increased accuracy when some systematics issues are resolved. Wuppertal measures the effect of source masses on the spacing of a pair of suspended masses which form a microwave Fabry-Perot cavity, and aims for accuracy better than 100 ppm. JILA uses its free-fall gravimeter to measure the change in g produced by a movable tungsten ring mass. BIPM uses a torsion balance suspended by a thin flat metal strip; its dominant torsional restoring force is gravitational, thus minimizing anelastic dangers. BIPM plans to use its instrument in two ways to measure G, in both a static displacement mode and a "time of swing" dynamic mode, aiming for a solidly reliable measurement at a 100 ppm level. The Russian lab uses a torsion pendulum in the classic dynamic mode used by Luther and Towler; it has been troubled in the past by poorly understood drifts.

Work in progress was reported by additional groups: Luther at LANL is developing an instrument which will use a bifilar suspension which, like the BIPM strip suspension, has a restoring torque which is dominantly gravitational in origin thus circumventing anelasticity issues. The University of Washington and Irvine labs are both building instruments which use thin plate pendulums suspended in nearly pure quadrupole gravitational field gradients produced by special source mass configurations. The Washington approach elegantly avoids fiber-related problems by servoing its pendulum to a continuously rotating platform whose measured periodic angular acceleration reflects that of the pendulum, while the pendulum fiber never twists significantly. The Irvine instrument uses the classic "time of swing" dynamic method, operating at 2K with a high Q fiber whose anelastic effects should be sufficiently small and well understood to not limit the measurement's accuracy. Both the JILA group and the Taiwan group of W.T. Ni discussed plans for G measurements using a scheme like Wuppertal's but using an optical rather than microwave interferometer as distance gauge. The SEE project was presented, which hopes to determine G and test other aspects of Newtonian gravity using measurements of ``horseshoe" trajectories of test masses projected toward a field mass within a long cylindrical space capsule. Development at the Politecnico di Torino of a G measurement using a pendulum swinging between two mass spheres was discussed.

The conference also featured fascinating talks by G. Gillies and I. Falconer on the history of G measurements and the painfully shy but highly skilled man Cavendish. Thibault Damour lectured on the theoretical importance of measurements of G and its possible dependencies on mass composition, distance, and time. Thibault reminded us that, contrary to popular belief, G is NOT the least well known fundamental constant - that distinction belongs to the strong coupling constant!

G measurement lab contacts: New Zealand: t.armstrong@irl.cri.nz Zurich: schlammi@physik.unizh.ch Wuppertal: meyer@wpos7.physik.uni-wuppertal.de, JILA: Fallerj@jila.colorado.edu BIPM: tquinn@si.bipm.fr Karagioz (Russia): irtrib@cityline.ru Ni (Taiwan): wtni@phys.nthu.edu.tw SEE: ASanders@utk.edu Politecnico di Torino: demarchi@polito.it Irvine: rdnewman@uci.edu Washington: gundlach@dan.npl.washington.edu


Jorge Pullin
1999-02-02