The data from the collimated sensors of the LEND instrument are shown to be of exceptionally high quality. Counting uncertainties are about 0.3% relative and are shown to be the only significant source of random error, thus conclusions based on small differences in count rates are valid. By comparison with the topography of Shoemaker crater, the spatial resolution of the instrument is shown to be consistent with the design value of 5 km for the radius of the circle over which half the counts from the lunar surface would be determined. The observed epithermal-neutron suppression factor due to the hydrogen deposit in Shoemaker crater of 0.25 ± 0.04 cps is consistent with the collimated field-of-view rate of 1.7 cps estimated by Mitrofanov et al. (2010a). The statistical significance of the neutron suppressed regions (NSRs) relative to the larger surrounding polar region is demonstrated, and it is shown that they are not closely related to the permanently shadowed regions. There is a significant increase in H content in the polar regions independent of the H content of the NSRs. The non-NSR H content increases directly with latitude, and the rate of increase is virtually identical at both poles. There is little or no increase with latitude outside the polar region. Various mechanisms to explain this steep increase in the non-NSR polar H with latitude are investigated, and it is suggested that thermal volatilization is responsible for the increase because it is minimized at the low surface temperatures close to the poles.