When we try to define the lower edge with the first collimator, we take a big hit because of the finite target length effects which prompted us to move the collimator downstream in the first place. Figure 1 shows how the profile barely clears the QTOR supports with the lower edge defined upstream, and even so the error on Qweak is 4.55%. Figure 2 shows the clearance of the profile for the current collimator (VT row 20). The finite target length effect is most visible in a plot of the unweighted theta_o acceptance on the cerenkov bars. Figure 3 shows that the extreme angles being essentially the same for the upstream edge definition (red) severely limits our acceptance from low theta as compared to the current downstream version of the collimator (black).


*All of these results are with air.

Version	Rate (MHz)	< Q 2>	error on Qweak (%)	bkgd from ep &gamma (%)	bkgd from mol e- (%)	bkgd from mol &gamma (%)
original VT row 20	816	.02611	4.27	0.203+/-.005	1.369+/-.151	0.664+/-.044
lower edge defined in 1st coll.	638	.02740	4.55	0.161+/-.005	2.828+/-.152	0.502+/-.029