The first page uses (except for the lower right-hand plot) 3-hit stubs with layer 0 missing. These are predominantly due to tracks that hit the front of the chamber--and thus have a very well-defined Z. The fits to rphi have a sigma of 4.11 (.36) and 4.61 (.61) for west and east respectively, and presumably the error in z should be similar, scaled by 1/sin(theta) (about 2.2), or 10. The fit numbers (left hand column) are 13.1 (3.6) and 8.4 (1.3) for west and east respectively--in the ballpark. However the fit requires 2 gaussians, with most of the events in the wider one.
The second page addresses the issue of these two gaussians. As you go across the page from left to right the sample becomes more and more restricted--but the ratio of the narrow to wide gaussians doesn't change substantially. The wide peak doesn't seem to be due to low quality tracks. However, it was proposed at the Monday meeting that this might be because the track came from a different vertex from that considered to be the primary vertex.
Back to the first page, and the right-hand column. For these 3-hit very-forward tracks, he plots the difference between the shadow time and the track time, and gets a peak centered at 1.63 (0.06) nsec and with a width of 1.42 (.054) nsec. Since this combines two hits, the resolution should therefore be (1.42/sqrt(2)) 1 nsec. The center is not completely understood yet. 1.3nsec of it is clear enough--the geometry of the cell (dead space and where these skimming tracks will hit) and the length of the jumper cable account for that much. When he compares (for good 4-hit tracks, scintillator hits required for quality) the difference between the outer and inner time differences, correcting for the slope, he gets the lower right plot, with a different effective resolution (1.2 nsec).
The third page is concerned with the comparison of our estimate of the Z of a stub, based on the difference between the shadow hit time and the prompt hit time, with that found from the COT/SVX. (In the top center plot, the large peaks at each side are overflow bins.) The bottom left plot is for Z's and is very clean. The fit is a cubic: (a*(t-.9)^3+b*(t-.9)+c). The deviation from the fit is shown in the upper right plot, which has a center at -4 and a sigma of 21.4 (3.6) cm. We think a 3-sigma cut of about 65cm for the deviation of our Z and the COT Z will be good-- though first we have to implement that cubic fit in the reconstruction. The lower right plot is for all stubs with Pt greater than 10, and it looks pretty lousy. The vertical band is probably accidentals.
Modified 4-August-2003 at 12:00
http://hep.physics.wisc.edu/~jnb/imu/6Aug2003