23/13/17  Comments while using SDP 
Running on working-12-07-17.smp.
PCall-3.out on all samplists X2red=63.00, loose DC=16.8, tight dXH3=1.20,tight SC=2.59
dXH3 & DC compete. PCall-4.out X2r=63.2 DC=17.1, dXH3=0.90, SC=2.56
Look at errs. R61 h9-12 look high.  Exclude both R61 and R62, X2r=80.00, CTFx looks great.  
Exclude samplists F62 and F61 in favor of the pairwise fixed angle samplists
 Call these RF***.out  Nomenclature R*** is rotation only and f*** is all pairwise rotations.
12/14  Look at RF61-a.out, only D=60.8. R61 h=9-12 off the fit. Profiles exceed limits - abb. as PEL = 0.1.
Fit only R61.  PEL->0.18   Fit all pairwise f61-a, PEL=0.1
The R61 higher orders were all significantly greater than the best fit but this was because the estsimated uncertainties were much tighter for the f61 data, so I increased the f7,9 and also the f5.7 uncertainties.  Now the fit runs closer to F61. All parameters look fine, except PEL~0.2.
From now on, the new working-c.smp will be used with *-b.out.
clicked off RF61 and put on RF62.  Decent agreement just with scaling. Same issue comparing h=9-13 with R62 larger due to tight constraints on f62.  Modified errors for f62 F5,7 and F7,9.
Rf62-b.out PEL~0.05, dXH2 too small, RCG close to target 0.42.  Decent fit.
Rfboth-b.out PEL~0.09 due to CH3, nearly the same as previous line otherwise. No significant improvement in PEL when r_t relaxed to 0.05.
Can make PEL=0 at CH3 by forcing larger Sc3~1.8, but fit much worse for high h.  Rfboth-c.out  Also, tightened dXH3, but SC ~ 3.1
Constrain SC=2.4+-0.1 --> 2.62,dXH3=1.2t0.005->0.74, SC==1.8, no PEL, but X2r=58.8. Rfboth-d.out actually smaller X2red than Rfboth-c.Erased.
Long hiatus until 1/2/18. Start LowOrders.smp for comparison of low orders from capillary versus new data. 
1/8  Have added estimated F1,2_1, F1,2_2 and F2,3_2 which had OE pixels. assigned large uncertainties. Using .smp with all new data plus capillary samples. TW and MW have larger h=1,2. Leaving those out gives a nice fit C3MP39.out. CMUcap632 from 2008 increases chi2 from 41 to 52. Adding TW&MW AllW67 also gives nice profiles. Now play with constraints. 
Naming conventions: C3mP30.out where C3 means three cap samplists, m means the methyl width is free, P means the Ph width was fixed and 30 is the rounded chi2. C3MP39 means the methyl width was fixed such that there is no excessive methyl probability. Freeing the methyl width sc3 is a big improvement to F(q) fit, much more so than freeing SPh. C3mp31 has 0.4 more chi2 than C3mP30 due to smaller penalties.  Negligible issue.
For some time it has been clear that SCh has to be constrained from shrinking to 0.
C5MP38 adds the TW & MW cap samplists to the two Rotations and the 17 Fixed angle samplists. Higher X2 but fits look very similar.
F17mp fits only the Fixed angle data. It needs data added for the h=8 zero