It isn't possible to tell if you are graphing consistently & accurately by looking at those graphs, because it is quite common for a gate's signature to look different to what you see around 56 on you last graphs. There is a great description of how gates can appear here, under the heading "failure to recognise a good indication" :
http://www.safeventures.com/news.php?news_id=16It would be a good idea to do a new AWL or AWR graph of the lock to compare to your existing graphs, because it will be a good indication of whether you are taking consistent readings. Once you are able to run a graph twice & get the same results, it really helps you to put faith in your readings.
With respect to learning by graphing one, two and then all three wheels, I don't think this would be very helpful.
If you have all gates on the same number, you won't get a gate signature on your graph, the fence will drop in when you get to the gates & the lock will open.
When you are graphing with two wheels set to the same number, you could actually complicate things, as high low tests will give strange results & the effect of rotational conversion will negate the fact that two wheels are set to the same number - if wheels 2 & 3 are both set to 50, you won't have both wheels under the fence if you dial both wheels to 50 in the same direction, W2@L50 W3@L50 is different to W2@R50 W3@L50.
Once you can take consistent readings, the skill of manipulation lies in interpreting graphs, deciding what to do next based upon your graphs and past experience & knowing how to move the wheels to accomplish your next step - you really need all three wheels in play to do this.
One common way to make things easier whilst learning is to bend the fence slightly, so that you know the lock will read wheel 3 first, then wheel 2 & finally wheel 1 . Bending the fence also makes the graphs a lot easier to read.
Hope that helps,
...Mark