“…Referring to the caesium relative decay rates shown in figure 7, we see that of this 35% loss, 5 24 (7%) will decay straight to F = 3, m F = 0 and stay there as incoherent background; 3 8 (13%) will decay to F = 3, m F = 1, 2 and not affect the interferometer, other than a drop in total signal; 5 12 (15%) will decay to various F = 4 states and get pumped around again, possibly resulting in another few per cent ending up in F = 3 or 4, m F = 0. The net result of all this optical pumping will be that after one 'out and straight back' AT, ∼60% of the initial atoms are returned to F = 4, m F = 0 coherently, ∼10% will be present as incoherent background in F = 3 or 4, m F = 0 and the rest will be lost from the states involved in the interferometer.…”