
Hi Bill,
Many thanks for manually slaving over my numbers with graph paper in hand. I owe you, let me know if you ever come to Ireland.
I wonder if you used 0.1 as the speedpoint, or WBM recommended 0.17 in arriving at the CIs. I am trying to understand the difference in the calculations, I suppose I should check with Ralph, too.
When I plugged your, much nicer I must admit, numbers back to the same WBM spreadsheet, I figured that my N is 6 min, N+1 8.5, N1 5. Also, if I use the shortcut of EI 200 for N, I get an easy to remember sequence of EI 150 for N1, 200 N, 300 N+1. However, I am not sure if I can use the spreadsheet for calculating those times if I use your kindly provided CIs, perhaps I need to arrive at them differently.
I am also not sure what CI to consider normal, as this is the first time I applied a densitometer to the job. I was planning to follow WBM 0.57 or nearby, or to do a paper test.
Thanks icracer for the contact exposure suggestions.
Many thanks for helping me. I've just arrived in Denver, setting off to WY in two days, armed with a new table of EIs. Feels like fun.

Hi Rafal,
Thanks, it's nothing...
Here is the graph.
http://www.beefalobill.com/images/rafaltxp.jpg
Shows how difficult curve fitting can be, where the goal is to find one number to represent the curve.
The "c" marks I drew on the curves are the points I used to give Contrast Index (CI) estimates.
Small diamonds are the 0.10 "speed" point (small diamonds to the right and above are the corresponding ASA triangle aim points). The larger double diamonds (kind of looks like a dream catcher) emphasize that you exactly hit the ASA triangle at 11 minutes.
Drawing a straight line, ASA triangle is 0.62 CI. But ASA aims for different points on the curve than CI. The 11 minutes curve is 0.75 CI by my estimation. Crazy. Shows how badly a single number fits the curves.
Also shows how much tolerance there is. A good argument can be made to choose anywhere from 7 to 11 minutes as Normal. I think 6 is a bit short side but have to agree it's close.

My reciprocity chart for TriX 320 4x5 says that a 1 second exposure should be corrected to 1.5 seconds, a 50% or halfstop difference. If your shutter gives you 0.9 seconds actual exposure at the 1 second setting, you're even more underexposed.
The underexposure aside, your contrast is greater than you would expect. I'd agree you're agitating too much.
Film speed is highly dependent on a number of variables. Meter, metering, exposure, flare, scene contrast, developer, agitation, temperature, color temperature of the light, etc. Starting to test at box speed is reasonable, but don't expect your results to match. My personal exposure for 320 TriX is 160. Yours likely will be different, but it will be yours, and once you've pinned down the variables and start to get consistent results, you'll be on your way. Film testing is just that  testing. One test does not indicate a trend. It gives you a starting point from which to compare more data.
Peter Gomena

Peter Gomena,
You have a valid point about the time.
But because the image is a step wedge, in terms of "underexposure", it doesn't matter because you can see the curves rise up out of the toe and and reach practical exposure levels.
Rafal,
You have basically created a reciprocity failure study at 0.9 or 1 seconds. You should shorten the time a bit, unless you decide that the reciprocity failure at this time is reasonably close to what you normally shoot.

Let’s be careful not to confuse exposure with film speed. Metering or the scene luminance range has nothing to do with the speed of film, although the meter can play an important part with certain methodologies toward determining “film speed.”
This brings me to another thought about film speed and that concerns expressing a film speed result without referencing the testing methodology used. We all know that different approaches yield different results. This is the strongest argument for standardization. When the ISO prefix is placed in front of a film speed, it is stating the procedures of the standard have been adhered to.
Comparing the WBM method and the ISO method clearly illustrates my point. Without getting into the merits of either method, there is around a 0.15 to 0.23 logH difference between the two speed points. The same test wedge with the same data will yield speeds differing between ˝ to 2/3 stops. So, if communicating speeds results is to have any validity, I thing the testing methodology needs to be included.

Bill 
Yes, you are correct about the step wedge readings and exposure, it's kind of a relative thing because you can see the information in the graphed data. Metering and exposure in the field is more critical and demands consistent methods if you're going to get consistent results, and it's a whole lot more challenging out there than it is in the darkroom! Whether a person's exposure index for a film matches box speed, or is half or twice box speed is irrelevant. The amount of light hitting the film needs to be correct and consistent regardless, and consistent testing will support good results.
Peter Gomena

Bill, the problem with having the CI 0.75 curve fit the ISO triangle is that it is way too high. Long toed curves do require a slightly higher CI to fit the ISO triangle, but not this much. So, what is happening here? The testing under the ISO standard and the determination of the ISO triangle is done under no flare conditions. Rafal's test consisted of sticking the step tablet up against the window and shooting it with his camera (optical system). He emphasized the need for testing under daylight conditions but missed that such a setup will produce flare. As we know flare compresses the shadow area. ISO speed is determined using the gradient of the shadow. CI is determined more from a fuller range of the film. In order to raise the gradient of the compressed shadows sufficient enough to fit into the ISO triangle, the CI will be excessive like we see with Rafal's examples.
What is happening here is that a nonflare interpretation is being used with a test that incorporated flare. Any results are questionable at best.

You are right, Stephen.
The ASA triangle is undefined in this set and must be discarded. And a good thing too, it feels wrong.
Just guessing, I think the development time that would fit ASA triangle (suppose Rafal makes a contact step wedge) might be a lot closer to 8 minutes.
The toes on all these curves show what you would draw as dotted lines (to show flare) on a contacted step wedge test set.
Since this test includes flare, it's good to visualize placement of scene values and predicting densities on negative. And it is certainly good enough for process control.
The CI values are not truly what I quoted (they might be "close), and the ASA triangle needs to go.

1 Attachment(s)
And not just the triangle either. All of the various gradient measurements are based on no flare curves  the gradient for contrast index or the base point for average gradient. Then there's the subject of speed determination with a flare curve vs a non flare curve. This is an excellent example of the importance of contacting step tablets.
I've superimposed a contacted curve of TXT in Xtol 1:1 over the curves you generated. The CI is 0.64 using a modified version of CI.
Attachment 56281

I think we need to "back out" the flare, perhaps with 0.4 assumption...