We use Subaru data to conduct a detailed weak-lensing study of the
dark matter distribution in a sample of 30 X-ray luminous galaxy
clusters at 0.15<z<0.3. A weak-lensing signal is detected at high
statistical significance in each cluster, the total detection S/N
ranging from 5 to 13. In this paper we concentrate on fitting
spherical models to the tangential distortion profiles of the
clusters. When the models are fitted to the clusters individually, we
are unable to discriminate statistically between SIS and NFW
models. However when the tangential distortion profiles of the
individual clusters are combined, and models fitted to the stacked
profile, the SIS model is rejected at 6- and 11-sigma, respectively,
for low- and high-mass bins. We also use the individual cluster NFW
model fits to investigate the relationship between cluster mass
(M_vir) and concentration (c_vir), finding an anti-correlation of
c_vir and M_vir. The best-fit c_vir-M_vir relation is: c_vir(M_vir)
propto M_vir^{-alpha} with alpha=0.41+/-0.19 -- i.e. a non-zero slope
is detected at 2sigma significance. We then investigate the optimal
radius within which to measure cluster mass, finding that the typical
fractional errors are improved to sigma(M_Delta)/M_Delta ~ 0.1-0.2 for
cluster masses at higher over-densities Delta=500-2000, from 0.2-0.3
for the virial over-density (~110). Further comparisons between mass
measurements based on spherical model fitting and the
model-independent aperture mass method reveal that the 2D aperture
mass enclosed within a cylinder of a given aperture radius is
systematically greater than the 3D spherical mass obtained from NFW
model fitting: M_2D/M_3D= 1.34 and 1.40 for Delta=500 and 110,
respectively. The amplitude of this effect agrees well with that
predicted by integrating the NFW model along the line-of-sight.