We utilise a two-color Lyman-Break selection criterion to search for z \sim 9 -10 galaxies over the first 19 clusters in the CLASH program .
A systematic search yields three z \sim 9 -10 candidates .
While we have already reported the most robust of these candidates , MACS1149-JD , two additional z \sim 9 candidates are also found and have H _ { 160 } -band magnitudes of \sim 26.2-26.9 .
A careful assessment of various sources of contamination suggests \lesssim 1 contaminants for our z \sim 9 -10 selection .
To determine the implications of these search results for the LF and SFR density at z \sim 9 , we introduce a new differential approach to deriving these quantities in lensing fields .
Our procedure is to derive the evolution by comparing the number of z \sim 9 -10 galaxy candidates found in CLASH with the number of galaxies in a slightly lower redshift sample ( after correcting for the differences in selection volumes ) , here taken to be z \sim 8 .
This procedure takes advantage of the fact that the relative volumes available for the z \sim 8 and z \sim 9 -10 selections behind lensing clusters are not greatly dependent on the details of the lensing models .
We find that the normalization of the UV LF at z \sim 9 is just 0.28 _ { -0.20 } ^ { +0.39 } \times that at z \sim 8 , \sim 1.4 _ { -0.8 } ^ { +3.0 } \times lower than extrapolating z \sim 4 -8 LF results .
While consistent with the evolution in the UV LF seen at z \sim 4 -8 , these results marginally favor a more rapid evolution at z > 8 .
Compared to similar evolutionary findings from the HUDF , our result is less insensitive to large-scale structure uncertainties , given our many independent sightlines on the high-redshift universe .