Cec,
behind all pattern-based elmination is some logic to explain it. You could use the pattern without actually dwelling on "why & how". Occasionally you have to answer these questions to grasp the main concept so that identifying patterns could become free of "Am I making a mistake here ?".
One way to make eliminations more logical is to Assume that your target elimination is actually a placement & see what are consequences of that assumption.
On all of these occasions it will lead to a situation of invalidity where a sector has no particular candidate or an x number of cells will have only an (x-1) number of different candidtaes.
This goes accross the board.
this has been exploited to combine information from groups of cells (that would otherwise sit there doing nothing) with other groups of cells in a similar situation to make a pattern that targets cells providing this invalidity if they turn out to be true. We could then move on & eliminate them.
The trick is to know how to combine the information from these different groups of cells. I'm using the term "sticky ends" to describe a way to do that. 1 sticky end from one group sticks on another from a different group.
in combining ALSs, this sticky end takes the form of a common candidate most of the time. there is a catch however. Each sticky end must stick to another sticky end to work. you can't leave a sticky end FREE.
now you could stick 2 groups of cells only or you could find another group of cells and STICK them to your CHAIN of groups.
At the end you'll have a curly bendy chain of cells with a FREE end at the START & a FREE end at the END and sticky ends between the groups inside it that used to join the information.
These FREE ENDS have now gained enough power to eliminate candidtaes.
Examples would make things even clearer.
tarek