weakness of the gravitational force makes it difficult to explore and characterize
all aspects of this interaction with great certainty. Therefore, we must
always consider the possibility that new gravitational effects can be discovered.
This is especially true since we donít have yet any unifying theory of
gravitation in the context of quantum physics, which could rule out some
hypothetical effects. Because of this, the last few decades have seen a
variety of claims about possible anomalous behavior of gravity as observed
during the course of a number of different kinds of laboratory investigations
One main line on these studies has been the question about spatial dependence of the gravitational constant G, i.e., is the inverse square law of the gravitational interaction correct or not, or in another words, are there undiscovered weak, long-range non-Newtonian gravitational forces. Even though convincing evidence of such a new force was never found this problem is highly important since it is directly linked to attempts to uncover violations of the weak equivalence principle of general relativity. Another widely investigated problem has been the temporal constancy of G. Also this question has great interest since measurements of temporal variations on G act as a significant test of various unification theories and cosmological models. Third long line in studies of gravity anomalies is searches for gravitational absorption and gravity shielding phenomena. These experiments have been motivated by the idea of the gravitational equivalence of magnetic permeability, an analog from electrodynamics. Common to all these experiments and other searches for gravity anomalies is that they all test the fundamental aspects of general relativity, quantum physics and cosmology, the corner stones of our understanding on physical reality.
An interesting set of unexplained observations of gravity anomalies has been found on experiments carried out during solar eclipses. Most of these studies have been performed using paraconical pendulums, torsion pendulums or spring gravimeters. In the experiments with paraconical pendulums (a modified version of Foucaltís pendulum, whose articulation consists in the contact between two planes via a sphere) the focus was on the orientation of the pendulum oscillation plane. During the solar eclipse in 1954 (at Le Bourget, France) Allais [2-4] found that the oscillation plane turned several degrees when the eclipse started (the first contact of the moon), stayed on this angle, and returned back at the end of the eclipse (the last contact of the moon). Similar type of anomalies was also reported by Savrov in eclipses in 1961 in Moscow , 1991 in Mexico City [6,7], and 1994 in Brazil , although effects were close to the limit of the instrument resolution.
Saxl and Allen  observed significant changes in the oscillation period of their torsion pendulum in the total solar eclipse in 1970. The period suddenly increased at the onset of the eclipse but curiously it remained on that level after eclipse. They also observed rather regular cyclic perturbation on the oscillation period well before the eclipse. Experiments with a torsion pendulum during the eclipse in 1990 in Finland made by Kuusela  didnít show any significant modulation on the period, which can be related to the temporal phase of the eclipse. The same eclipse was also observed in Bielemorks with a torsion pendulum , and in Finland with an absolute gravimeter  and water tube clinometers  but results were negative. New measurements with the improved torsion pendulum by Kuusela  in 1991 in Mexico City again gave negative results when concerning the pendulum period but interestingly the position of the pendulum wire in the horizontal plane encountered shifts which seemed to be related to the onset and end of the eclipse, as seen in Fig. 1. The effect was small but statistically significant.
|Figure 1. The x and y position of the torsion pendulum wire in the horizontal plane as a function of time. The first contact is marked with a, the maximum with b and the last contact with c. The shifts in the y position are marked with arrows (Kuusela T. Phys. Rev. D 43: 2041-2043,1991)|
|The latest positive results on gravity anomalies during the total solar eclipse have been observed with zero-length spring gravimeters; these instruments measure directly the vertical component of the gravitation field. In 1995 in India clear decrease of 10-12 mgal (1 mgal = 10-8 m/s2) in the gravity field was found which occurred at the onset of the solar eclipse . The most striking gravimeter recording was done in Moho, China in 1997 solar eclipse [16,17]. In this study there were two time regions with significant decrease in the gravity field. The first one occurred within 30 minutes before the first contact and the second one also within 30 minutes after the last contact (see Fig. 2).|
|Figure 2. Variations of vertical gravity measured during the total solar eclipse (Wang QS et al. Phys. Rev. D 62: 041101, 2000).|
drop in the gravity field was about 7 mgal,
2 or 3 times the noise amplitude. It should be noticed that timing of these
drops in the vertical gravity component are rather similar than the shifts
of the torsion pendulum wire in the horizontal plane shown in Fig. 1. Also
in all other positive results the change was always related to the onset
and end of the eclipse, not to the short moment of totality.
All gravity anomalies observed during solar eclipse can be explained by the tilt of the apparent vertical direction, either because of the tilt of the instrument (or the basis under instrument) itself, rotation of the gravity vector or generation of an additional horizontal component on the gravity field. Unfortunately the vertical direction has not been normally measured during eclipse experiments with pendulums except in the experiment in 1991 by Kuusela  where the shifts in the position of the pendulum wire were direct indications of change in vertical direction. Especially paraconical pendulums are very prone to any tilt if the azimuth of the oscillation plane is only measured from one side with respect to the usual pendulum rest position. In torsion pendulums the tilt produces significant effects on the oscillation period if the period of the half cycle is measured as it was done in the experiments of Sax and Allen . When the total cycle is measured, the effect of the tilt is much smaller, and typically no change in the period can be observed even if there are small but evident changes in apparent vertical direction, as demonstrated by Kuusela . Since gravimeters measure the vertical component of the gravity field, any tilt produces drop in the apparent value of the gravity although gravimeters cannot directly detect the horizontal component. We can conclude that the unexplained observations in 1991 eclipse could be now understood when comparing them with the new ones made by gravimeters.
have planned a solar eclipse expedition using a new approach. First, we
will directly measure the horizontal component of the gravity field during
solar eclipse, first time in the history of these kinds of experiments.
Secondly, we will perform observations simultaneously in distant locations
within the totality path with equal instruments. Single positive results
reported in past years has an evident weakness: the validity of the observation
is not possible to estimate in lack of any comparative data. The use of
equal instrumentation in several places is also essential since then we
can really explore temporal and spatial correlations of the possible gravity
Gilles GT. Rep. Prog. Phys. 60, 151-225 (1997)