My group seeks highly accurate 3J 1H-1H coupling constants. After an extensive literature search, I see conflicting statements about zero filling in 1D 1H NMR and whether it... a) improves the resolution of a spectrum? b) improves the accuracy with which the peak position - the top of the peak - is known? c) simply uses the added zeros to interpolate between the experimental data points, and does this improve accuracy of the 3J 1H-1H coupling constants? d) since the FT "throws out" half the data, does one zero filling (up to the original number of data points acquired) actually help?