By a strange sequence of actions, I noticed that Christian Peckham has produced a very clever variation of the Cyclic Four quarter peal.

I was searching CompLib for a quarter peal composition of Elliptic Jump Major, using the "extended search" feature which tests all of the stored compositions against a specified method. This produced all sorts of possibilities of different lengths, one of which was derived from a composition of Albany Little Treble Place Major (which is one of the Cyclic Four methods) by Robert Brown. I hadn't seen that before, but it turns out that he has produced a few peals and quarters for it. I then looked up Albany and clicked on the link for "compositions designed for this method", and one of them was this quarter by Christian.

It's a cyclic 8-part with the following calling.

1464 Spliced Major (4m)
Simon J. Gay and Christian M. Peckham
          12345678
------------------
Caterham  14263857
Albany    13527486
Hull      16482735
Hull      15738264
Albany    18674523
Caterham  17856342
Huntlink  23456781
------------------
8 part.

This is the same as the original Cyclic Four, except that Bristol is replaced by Caterham, up to the lead end 17856342. Then comes a different link method, Huntlink, which is plain hunting on eight but (and I had to look quite closely for a while before I realised this) only for seven changes! Like this, with handstrokes and backstrokes marked:

B 17856342
H 71583624
B 75138264
H 57312846
B 53721486
H 35274168
B 32547618
H 23456781

and then the next part starts at backstroke, so the methods are rung with all the handstrokes and backstrokes reversed. Very difficult, I expect, but it would be interesting to try it.

The reason why Bristol has to be replaced by a different method is that the first part end, 23456781, occurs at the handstroke of the first half lead of the plain course. Easier methods such as Norwich, Kent, Oxford are also false, for more complicated reasons I suppose; evidently Christian looked into all this and came up with Caterham, which is one of the methods in Chandler's 23-spliced and therefore somewhat standard. He's put another version on CompLib with a method called Timothy Treble Bob which is Oxford with Kent places in 5-6 before and after the lead end; but anyway, Caterham is Glasgow above the treble and therefore perhaps within reach.

After digesting what's going on in this composition, I remembered seeing a similar idea before, in this date touch from last year with a composition by Don Morrison:

2,024 Spliced Major (9 methods)
Donald F Morrison
  12345678  Cambridge Surprise
  15738264  Double Norwich Court Bob
  14263857  London Surprise
  16482735  (link)
  61428375  Wellington Little Bob
- 68753421  Yorkshire Surprise
  63271845  Lessness Surprise
- 68732514  Superlative Surprise
- 67214853  Cornwall Surprise
- 68172345  Bristol Surprise
- 67812345
Repeat seven times.
The (link) is one change with place notation 38.
Note that half the time the ringing is on the “wrong” stroke.
Contains 256 each Bristol Surprise, Cambridge Surprise, Cornwall Surprise, Lessness Surprise, London Surprise, Superlative Surprise and Yorkshire Surprise, 128 Double Norwich Court Bob and 96 Wellington Little Bob, with 71 changes of method, all the work of every method for every bell (including the treble), and 90 runs of four or more consecutive bells either at the back (65) or at the front (25).

The methods are the Core Seven plus Double Norwich and Wellington Little Bob. The latter is like Little Bob but with the treble going to 6th place and plain hunting at the lead end, presumably to get the desired length. The key feature of this composition though is the link "method" with an odd number of changes, in fact just one change with the place notation 38. As the notes say, this means that the next method starts at backstroke. I must say that I like the elegance of Christian's composition with the link method at the end of the part, but I don't know which link method would be easier to ring.

Christian has another version with a slightly shorter link method that moreover has an even number of changes per lead and doesn't reverse the strokes. The link method is more complicated though, with place notation 123678x1458x:

17856342
17865342
71683524
76183254
67812345

I'm now in danger of accumulating 8-bell projects faster than we can ring them, but let's see what we can do.