We investigate the injection spectrum of ultra-high-energy (>10(15) eV
) cosmic rays under the hypotheses that (1) these cosmic rays are prot
ons and (2) the sources of these cosmic rays are extragalactic and are
distributed homogeneously in space, although they may have had a diff
erent strength in the past; furthermore, we assume that we are not unu
sually close to any individual source(s). The most puzzling aspect of
the observed ultra-high-energy cosmic-ray spectrum is the apparent non
existence of a ''Greisen cutoff'' at about 10(19.8) eV. Such a cutoff
would be expected owing to rapid energy loss from photopion production
caused by interactions with the microwave background. We show that th
is fact could be explained naturally if most (or all) of the cosmic ra
ys presently observed above about 10(19.6) eV were initially injected
with energy above the Greisen cutoff. However, we find that the inject
ion of cosmic rays above the Greisen cutoff cannot account for the obs
erved flux below about 10(19.6) eV Unless the injection rate of these
particles was enormously higher in the past, as would be the case if t
he injection resulted from the decay of an ultramassive particle with
lifetime of order 10(9) yr. Even with such a rapid source evolution, t
he observed cosmic-ray spectrum below about 10(18.5) eV cannot be expl
ained by injection of particles above the Greisen cutoff in the distan
t past. However, we show that a 1/E(3) injection spectrum can account
for the observed spectrum below 10(18.5) eV, with the steepening obser
ved by the Fly's Eye group between 10(17.6) and 10(18.5) eV being very
naturally explained by e(+)e(-) production effects. This latter fact
lends support to the hypothesis that the cosmic rays in this energy re
gime are protons of extragalactic origin. However, owing to e(+)e(-) p
roduction effects, a 1/E(3) injection spectrum cannot account for the
observed flux above about 10(18.5) eV.