The Russian-American experiment SAGE began to measure the solar neutrino
capture rate with a target of gallium metal in Dec. 1989. Measurements have
continued with only a few brief interruptions since that time. We give here the
experimental improvements in SAGE since its last published data summary in Dec.
2001. Assuming the solar neutrino production rate was constant during the
period of data collection, combined analysis of 168 extractions through Dec.
2007 gives a capture rate of solar neutrinos with energy more than 233 keV of
65.4 (+3.1)(-3.0) (stat) (+2.6)(-2.8) (syst) SNU. The weighted average of the
results of all three Ga solar neutrino experiments, SAGE, Gallex, and GNO, is
now 66.1 +/- 3.1 SNU, where statistical and systematic uncertainties have been
combined in quadrature. During the recent period of data collection a new test
of SAGE was made with a reactor-produced 37Ar neutrino source. The ratio of
observed to calculated rates in this experiment, combined with the measured
rates in the three prior 51Cr neutrino-source experiments with Ga, is 0.87 +/-
0.05. A probable explanation for this low result is that the cross section for
neutrino capture by the two lowest-lying excited states in 71Ge has been
overestimated. If we assume these cross sections are zero, then the standard
solar model including neutrino oscillations predicts a total capture rate in Ga
in the range of 63-66 SNU with an uncertainty of about 4%, in good agreement
with experiment. We derive the current value of the neutrino flux produced in
the Sun by the proton-proton fusion reaction to be (6.0 +/- 0.8) x
10^(10)/(cm^2 s), which agrees well with the pp flux predicted by the standard
solar model. Finally, we show that the data are consistent with the assumption
that the solar neutrino production rate is constant in time.