We have determined the inhomogeneities in the density of prototype STEP tes
t masses made from HIPed beryllium and niobium using the hydrostatic weighi
ng method developed at BIPM. 13 samples are taken from a HIPed beryllium ro
d (approximately 140 mm long by 90 mm) and 24 samples from a niobium rod (a
pproximately 225 mm long by 45 mm). The maximum relative difference of dens
ity is found to be approximately 240 ppm in the beryllium samples and 60 pp
m in the niobium samples. Assuming that there is a 10 g helium bubble at 25
0 mm from the test masses acting as;a source, these inhomogeneities of dens
ity would produce a disturbance of Deltaa/a = (1.1 +/- 0.7) x 10(-18) in a
beryllium outer test mass and (5.1 +/- 0.5) x 10(-20) in a niobium inner te
st mass, in the STEP experiment. The inhomogeneities of density in the HIPe
d beryllium rod appear to be too large for the STEP experiment, which aims
at a sensitivity of one part in 10(18). We suggest that it is important to
find a way to fabricate more homogeneous HIPed beryllium.