by **Sue De Coq** » Sat Aug 27, 2005 5:41 am

The aforemention chain forms a pattern that is very similar to the Turbot Fish described by nick70. (I'm not sure whether it's exactly a Turbot Fish as the two non-consecutive strong sides, i.e. r9c6-r9c3 and r1c3-r1c5, aren't connected by a box with just two candidates, as described in the original post on the subject. However, the underlying logical justification is almost identical). So, to find a chain such as this, simply use a slightly-extended form of whatever algorithm you use to find the Turbot Fish/Swordfish/X-Wing patterns. I claim that my tactic should be considered legitimate (i.e. not T&E) if Turbot Fish, Swordfish and X-Wing are all considered legitimate.

Now that I've re-read nick70, I see that he prefers a shorter many-valued chain to a longer single-valued chain, in which case he might like:

r7c5=9 => r7c2<>9 => r7c2=8 => r3c2<>8 => r3c2=9 => r1c3<>9 => r1c5=9, which has just four legs. (Since r7c5=9 leads to a contradiction, it's eliminated, which forces the 9 in Column 5 into Box 2, so 9 is no longer a candidate for r2c6).