[r-t] Chris Kippin: "How my all-the-work peal of Spliced Surprise Royal were put together"

Fri Dec 22 11:46:24 UTC 2006

Hi people,

Earlier this year Chris Kippin very kindly supplied me with some detailed background information which I had requested on how he put together his well-known atw peals of spliced surprise royal.  It's very interesting, and when I recently re-read Chris's notes, I thought they deserved to be shared more widely. This morning I asked Chris if he was happy for me to distribute it on here, and he has replied positively.  

So consider it a Christmas present to you all! The text below is scanned in from the hard copy Chris sent me, so there may be some issues with the automatic text recognition.  I'll also attach the notes as a .pdf file shortly.

All the best,

Philip

How my all-the-work peal of Spliced Surprise Royal were put together

Firstly - why?

I had long been an admirer of the Pitman series of all-the-work Spliced S Major series from 4 to 9 methods and had rung and called some of them. Pitman had composed an all-the-work peal of 4-Spliced S Royal too, but he decided to retain an extending-lead method and used Rochester, a rather unsatisfactory method (though with a nice name), and a 5-part with Rochester and York. Derek Ogden had also produced an all the work 4-Spliced S Royal using Redcliffe, a much better extending lead method, though rather tricky. But of course using an extending lead method in Royal means that, unless you part the tenors, you have to ring at least 8 consecutive leads of it to get all the work for the back four. Pitman had also used this approach with York. So I felt there wasn't really a satisfactory all-the-work peal of Spliced S Royal around at that time, ie the early 1970s.

The catalyst

One evening I was asked to call a quarter of Spliced S Royal at Christ Church Bristol, in, I think, London No 3, Cambridge and Yorkshire. It occurred to me that if! called each course LXXLXXLXX, but rotated the calling in each course, all bells would ring all the work. But I soon also discovered that this arrangement, with three bobs in any position, is also very obviously false (eg you can get 2143658709 repeated, or 1324567890 and so on, which even ordinary ringers would notice!). So I wondered if it might be possible to use this type of arrangement to obtain all the work for the back four, and fill the gaps in for the front bells in other courses.

Which methods?

For the reasons stated above I didn't want to use an extending lead method. Bristol is an excellent Royal method which we used to ring a lot in Bristol and clearly would combine with leads orders b and f in quite different ways than mx would. The main feature which would be lost would be the extending lead, especially for blocks of 2 or 3 consecutive calls. For the rest of the methods I decided to use London No 3, Cambridge and Superlative No 2. Some people might object that the last is an odd method which has few redeeming features, but I don't agree. Its difference adds greatly to the interest of the peal and its musical qualities are there if you listen for them, particularly the 567890 which occurs 6 changes after the wrong, when the 38 place seems to be leading in a totally different direction. Beethoven would have liked this if he'd been a ringer! In addition, my father had called the first peals of London No 3, Bristol and Superlative No 2 S Royal, and the first peal of 4-Spliced S Royal, using those methods (3-lead courses, using bobbed leads of Bristol as b), all at Beddington in the 1930s, so it had to be those.

How did I get started?

I had no previous experience of composing Spliced Surprise so I had to start from first principles. I don't have the sort of brain that Tony Cox has, so some of my methods of working might seem long-winded, but I had to be able to understand them. And remember this was in 1973 or so - no PCs then, and only mainframe batch-processing proving programs running at places like Harwell and Bristol University (turnround time measured in days). So it all had to be done in a scientific manner, rather than trial and error, and so the first thing I had to do was to work out some proof scales. I did this in what will probably seem to you like an incredibly labour-intensive way, but it worked.

1          First I wrote out, on four sheets of graph paper, a whole course of each of the 4

            methods (1080 rows).

2          By the side of each row I wrote the position of the treble, the nature of the row and the

            position of 7, 8, 9 and 10 (eg 2-8709 for the 1 st row of Cambridge, 2+6980 for the

            first lead of London etc), also the method name and lead number.

3          Then I cut the four sheets into 1440 incredibly small slips, taking care not to lose any

            on the carpet, in my trouser turnups etc (the spell check prefers turnips).

4          Then I sorted them by treble position and nature (eg 1 +, 1- etc) and stored them in 20

            envelopes, one for each.

5          Then I took each envelope in turn and sorted the 72 slips by the position of7890.

6          Where there were duplicates I transposed the two rows to give me a false course head

against the plain. course and its incidence, eg London 1 is false against Cambridge 7 in   43526. I included the 23456 fces - not only those where one lead of a method is false against a different one in the plain course (eg B against S, at the half lead) but also the full lead ones (eg 23456 Cl is false against 23456 SI): this proved useful in saving me from very silly errors later on.

7          As I worked them out I wrote them on a set of 36 blank: computer punched cards, one colour for each lead head order (blue for g, buff for f and salmon for b!), adding to existing  fces where necessary. The cards were a brainwave, and were so easy to use later on compared with having the proof scales on sheets of A4 (or even foolscap!). They could be held like a pack of playing cards and used on trains and aeroplanes with ease. I still have them - they are a family heirloom.

All this took ages, and I can remember working very late one night listening to general election results coming in while I was doing it.

The first stages of the composition

Having got my proof scales I started to think about how I might get all the work. If you use three bobs in a Q-set (eg 3 homes) and use the LXXLXXLXX course, but rotate it the other way you get all the work for the back six but not for the front three. So that was where I started. I then thought about working in some Bristol (bit like making a cake really), so started on 52436 and 35426, another Q-set block, and constructed some courses using some leads of Bristol. It became clear fairly early on that some leads were pretty false and should be got in sooner rather than left till later, eg B9 went into 35426 where it remained. I also realised that there were relatively few course constructions available using only L and X, ie LXXLXXLXX

LXL

LXXXXX .

and their rotations, but adding B increased the number and interest significantly: LXBBXXX

LXXBBXX

LXBXXLXX

LXBXBXX

BL

BXXXX

BXBXBX

In addition a bobbed lead of B can take the place of a bobbed lead of X.

I went on in this sort of semi-haphazard way until I had used all the courses with 6 at home. Then I took stock.

Recording what I had done

For each course I worked out the various options available by proposing constructions, looking at each proposed lead in turn and seeing if it was true or false. There are a lot of potential blind alleys doing it like this, and I must have wandered up many hundreds. But I decided it was still quicker, and safer, than working out all the leads which were false once I'd used a lead and recording all of those. Unwinding that is also very tricky if you decide not to use a lead after all. So each course had a lot of scribble and crossings out, and a final set of leads used. I tried to keep my options open by using both C and S alternatively where possible, and only ditching one of them if something else later made that imperative. All of that was recorded on A4 sheets. At various points I rewrote these lists, as all I have now is a final list of the courses and the leads used.

I also needed to keep track of the all-the-work status, so once I'd got something like a round block I wrote out all the lead heads, with the calls and methods alongside, on large squared paper, so that I could then construct a grid of lead starts for each method. Everything was done in pencil so that it could be adjusted as I went along. I ditched these sheets once I'd completed the peals, as there didn't seem to be much point in keeping them (I have enough junk as it is - ask Heather!).

Extending the block

Once I'd exhausted the courses with the 6 at home (11 of them were used eventually, only 45236 ending up on the cutting room floor) I began to think about what else I wanted in the peal. The xxx65s were the next obvious courses, so I worked on those. There's quite a bit of xxx65 falseness, so they were a bit tricky. Then I started to add other courses to join them up, and, as you can see, the peal didn't all finish up as a set of Q-sets. I went on adding courses, building up the length and knocking off unfulfilled starts until I had the length and all the work. In the process I used 64523 which until that time no-one, as far as I know, had regarded as a musical course (though I am open to correction). It was only when we started ringing the peals that we realised what it contained! Marcus Sherwood, in his review of one of them, also expressed admiration for the 3 middles on 43526, keeping 456 coursing, but of course he's a cru man, something I find pretty insignificant on 8 and totally meaningless on 10 and 12! Once I'd got a workable peal I went on tinkering with it to try to achieve the following:

*	Perfect method balance - something AJP was keen on
*	Minimising the number of consecutive leads of methods. I had to accept, in the first version, 2 occurrences of 4 consecutive leads of B, but managed to get all methods down to a maximum of 2 in version 2.
*	Minimising the number of consecutive leads in which a particular method doesn't occur. I can't remember what the statistics were now, but I thought this was important for interest in ringing the compositions.
*	Maximising the 567890 and 657890 roll-ups. I am still cross, whenever I ring the 4¬Spliced, that I managed not to get 1234657890, and I think the defect is in the whole series.

The completed peal

Once I got to a stage where I couldn't improve it any further, or get fed up with trying, I gave it to Jim Taylor to prove at Bristol University. I still have a photocopy of the printout with Jim's comment 'True 27.11.74'). We rang it at Christ Church Bristol on 20/7/75, having met short earlier in the year.

The rest of the series

Once I'd completed the 4-Spliced I wanted to go on. I had discovered, whilst working on the four, that Jim Diserens had a computer program that would work out inter-method falseness, and the complete standard 8 Royal proof scales were at Jim Taylor's house, only 3 miles down the road. So I borrowed them and added them to the computer cards. This was another long task, but much quicker than working them all out by hand. I though initially of starting the 5-Spliced from scratch, but decided eventually that it would be easier to start with the 4 and adjust it. This wasn't too difficult, but of course I had to do more than just substitute Y for C and S in order to keep the methods in balance. Moving from 5 to 6 was more difficult still, as I had to add another f group method. If you look at the whole series you will see many similarities and many dissimilarities - in some the compositions appear entirely different shapes, but much of the basic material is still the same.

The second version of the four was interesting. Some time after it had appeared in the RW David Friend, then of Derby where they had a good ten bell band, asked me if the composition would be true if Yorkshire was substituted for Superlative. I said almost certainly not, but when I looked at it I found it was only false in a comparatively small number ofleads. I guess this was because S and Y have the same inter-lead 23456 falseness (ie they both have back rounds and so on). So I made some adjustments to get a composition where either S or Y could be rung throughout, and also got rid of the two lots of 4 consecutive leads of Bristol. This is the version which was published: the original version was dusted off for the 25th anniversary in 2000.

Later compositions

These consist of:

*	An alternative progression from 4 to 8 for John Mayne's handbell band, adding the methods in a different order. Never finished, because they only rang the first one. These are by and large simply redistributions of the methods in the original peal. Not published in the RW, but possibly the rung one in the Hertford County report.
*	5 and 6-Spliced atw, including Wootton Rivers (5) and Greenwich (6). Hard stuff it would appear - I think the 6 defeated Nick Simon' s band, although I might be wrong. I've rung the 5. Totally new compositions not related to the original series at all.
*	5-Spliced atw, LBCY plus Redcliffe, put together for John's handbell band, at Dorothea's request. Has a full course ofRedcliffe, necessarily, which I don't really like, but it was the only way to keep her quiet. There was a funny story to this. I normally like to get all compositions of this sort independently checked for truth and atw. I used to get Tom Chapman to do it, but after he died I asked Tony Cox for this one. Unfortunately he showed it to Peter Randall while he had it, with the result that Peter called it at Worsley before John's handbell band. John and Dorothea were furious, and submitted it with a very pointed footnote when they rang it: 'This composition, composed specially for this band, is now rung for the first time on handbells'!
*	A small series from 6 to 8 methods, LBCSY plus Kegworth ( a very musical method), Hillingdon and Kingussie, all methods with some wrong hunting but a good structure. Damned with faint praise when reviewed in the RW! All called by Derek Sibson.
*	A 10 atw, standard 8 plus Clyde and Lockington. Also has complete courses of the latter as I couldn't get (or haven't so far got) a set of courses which mixed them up and got atw. Called by DES but not I believe published
*	10000 standard 8 atw in 1 part: a simple extension of the basic peal. Produced for Jim Belshaw and the Kensington mob ages ago but never attempted, so far as I'm aware. Was given to SJLL as a Christmas present in 2004, an event which he dined out on several times in the year after. I think he wants to go for it silent and nc.
*	10560 LBCS atw in 2-parts, rung on the old 10 at Leighton Buzzard before the fire.
*	12120 LBCS atw in 2 parts, an extension of the above, rung on handbells by David Marshall et al.
*	20000 LBCSY, Nideggen and Jacob's Wells, rung at Lyme Regis and fully written up in the R W with the peal.
*	6-Spliced Delight Royal atw, for Roy le Marechal, which I'm waiting for him to ring.
*	6 and 8-Spliced in 2 parts, not atw, with 24 56s and 65s. Both rung.
*	4-21 Spliced, not atw, in 6 parts, like Norman Smith's Major. 10 and 12 rung in Bristol, 14 lost when Heather broke her bell and peal lying in a drawer ever since.
*	5000 25-Spliced in 5 parts, not atw, silver wedding present for Martin and Maggie Whiteley, and which Martin has been trying to arrange to ring ever since.
*	Lots of jottings etc: I've just found a 10560 in 3 parts in methods LSN and A. I haven't the slightest idea what the methods are or what is was composed for. As there's also an 8160 and a 9000 it looks like an attempt for a longer length which I gave up at that point. Also an attempt for a long length in 3 parts containing KMBSY Land W which seems to have got to 12600. I think M was Middlesex, and I think this was for Mark Regan, also for Leighton, which got stuffed after the bells went up in flames.

Further technical issues

After the original series I returned to the question of whether it was possible to get 3 courses of LXXLXXLXX with all the work of three methods in three courses. After a lot of playing around I discovered a set of courses which will do that, as far as I know the only one (assuming L starts 30 and X starts x30). If you look at some of my later peals you'll see what the group is! I also went on, much later, to see whether there was a group of9 courses in which 9 atw might be achieved in the same way, and that exists too, an extension of the previous idea. The 10 atw uses this group; including Bristol reduces the number of methods by one, but joining these disparate 8 courses together, and fitting in a couple of mx methods as well, was quite tricky. I'm not particularly pleased with the resulting peal- you lose a Jot of musical courses to do it and the complete courses of mx are, in my view, a blight. Maybe there's more work to be done here. Also the courses, being cyclic in terms of methods, tend to be rather similar, which I think is a drawback.

Software

As computers became more commonplace I began to think about how the process might be speeded up. When I visited David Beard in 1986 he showed me the output of a program running on a mainframe which calculated what leads were false/still available for any given input of leads, and displayed the results in the form of a grid. I went back home and wrote a program to do this on a PC, and it has proved most useful. It does nothing creative at all, simply does many times faster what I could do by hand. It takes a file of proof scales, output from a program written by Glenn Taylor, and then displays any required course, showing the methods down the side of the grid and the lead numbers across the top. Leads used show in green, leads false show in red and leads still available show in white.

When I was working on it I couldn't see how to wind the results back if! decided to abandon a lead and make it available again. If I calculated all the false leads which were no longer false against the abandoned lead, how could I know that some were not still false against something else? Glenn suggested I use a counter instead of an indicator - a simple yet brilliant idea. So the falseness is shown as a number - the larger the number the more occurrences of falseness are involved. If a lead is abandoned and an incidence of falseness is therefore released, the counter is reduced by one. Only when the counter gets to zero is the lead no longer false against anything. Simple, but I doubt I'd have thought of it myself.

I've since extended the program to work with peals in parts, though I know there's a deficiency in that part which I've never had time to sort out.

I've also written a separate program which keeps tabs on the all-the-work status of a peal in progress. Unfortunately it's a totally separate program, but I've never had the time or the inclination to try to combine them. They work in such different ways that the task would be enormous: it's easier to live with them as they are. I've also recently written a new program which displays all the changes in any lead of any method in any course so that I can see where the music comes - something I can't do very quickly on paper.

Future plans

Lots of idea, but no time until I've finished my degree! More methods might be a possibility, but I suspect that 8 or so is probably about as far as it's possible to go without making the peal too predictable or boring. Roddy Horton has achieved the ultimate - 14 atw in 5040, but, with all due respect, the peal is probably a greater technical achievement than interesting to ring. He constructs his peals on the basis of 14 courses which contain all the work, then fits the methods to the calls. He also makes extensive (if not exclusive) use of ABCABCABC rotated, where ABC don't give the sort of falseness that LXX does, and joins them in Q-sets. This often means that you ring ABC etc for 3 courses, then say goodbye and move on to something else. This isn't really my idea of Spliced - I think there should be some element of surprise/suspense for the band, if not for the conductor.

I'm more interested in more musical peals. I promised to produce a 5000 in the 7 methods we rang for the 20000, utilising some of the tenors-parted courses. And 4-Spliced in the standard methods with tenor-parted would be good too. And peals in different groups of methods, ones like Kegworth which have more musical possibilities than the standard ones. Then there's really musical stuff using designer methods, eg the sort of thing David Hull might produce if he was interested. All I need is the time...

Chris Kippin 

January 2006

DISCLAIMER:

This communication (including any attachments) is intended for the use of the addressee only and may contain confidential, privileged or copyright material. It may not be relied upon or disclosed to any other person without the consent of the RSC. If you have received it in error, please contact us immediately. Any advice given by the RSC has been carefully formulated but is necessarily based on the information available, and the RSC cannot be held responsible for accuracy or completeness. In this respect, the RSC owes no duty of care and shall not be liable for any resulting damage or loss. The RSC acknowledges that a disclaimer cannot restrict liability at law for personal injury or death arising through a finding of negligence. The RSC does not warrant that its emails or attachments are Virus-free: Please rely on your own screening.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://bellringers.net/pipermail/ringing-theory_bellringers.net/attachments/20061222/5c2c793a/attachment-0003.html>