<span style=" font-size:10pt;font-family:sans-serif">I have created a
collection on CompLib </span><a href=https://complib.org/collection/11125><span style=" font-size:10pt;color:blue;font-family:sans-serif">https://complib.org/collection/11125</span></a><span style=" font-size:10pt;font-family:sans-serif">
based on </span><a href=https://changeringing.co.uk/cambridgefm.htm><span style=" font-size:10pt;color:blue;font-family:sans-serif">https://changeringing.co.uk/cambridgefm.htm</span></a>
<br><span style=" font-size:10pt;font-family:sans-serif">I also went through
all the compositions on CompLib with the same characteristics:</span>
<br><span style=" font-size:10pt;font-family:sans-serif">Only Cambridge
Surprise Major</span>
<br><span style=" font-size:10pt;font-family:sans-serif">>= 5000 changes</span>
<br><span style=" font-size:10pt;font-family:sans-serif">Tenors together</span>
<br><span style=" font-size:10pt;font-family:sans-serif">Tenors strictly
together</span>
<br><span style=" font-size:10pt;font-family:sans-serif">Standard calls</span>
<br><span style=" font-size:10pt;font-family:sans-serif">Normal start</span>
<br>
<br><a href="https://complib.org/composition/search?backstrokestart=false&midleadstart=false&variablehunt=false&variablecover=false&standardcalls=true&tenorstogether=true&strictlytenorstogether=true&minimumlength=5000&nummethods=1&method-1-value=true&method-1-title=Cambridge%20Surprise%20Major&sort=length"><span style=" font-size:10pt;color:blue;font-family:sans-serif">https://complib.org/composition/search?backstrokestart=false&midleadstart=false&variablehunt=false&variablecover=false&standardcalls=true&tenorstogether=true&strictlytenorstogether=true&minimumlength=5000&nummethods=1&method-1-value=true&method-1-title=Cambridge%20Surprise%20Major&sort=length</span></a>
<br>
<br><span style=" font-size:10pt;font-family:sans-serif">and found 74 compositions.
Some of those compositions didn't seem to be small variations of the 48
representative compositions, so I added all the compositions to the collection,
and categorised the ones over the 48 as extra or similar. [I think the
48 were just grouped by parts, length and number of befores].</span>
<br>
<br><span style=" font-size:10pt;font-family:sans-serif">For similarity
I used the following:</span>
<br><span style=" font-size:10pt;font-family:sans-serif">Generate the rows
for each composition.</span>
<br><span style=" font-size:10pt;font-family:sans-serif">When comparing
a composition to a reference composition, chose the rotation / reversal
which maximises the number of row + next row pairs in the composition which
match a row + next row pair in the reference.</span>
<br><span style=" font-size:10pt;font-family:sans-serif">See </span><a href="https://lists.ringingworld.co.uk/pipermail/ringing-theory_bellringers.org/2019-February/026966.html"><span style=" font-size:10pt;color:blue;font-family:sans-serif">https://lists.ringingworld.co.uk/pipermail/ringing-theory_bellringers.org/2019-February/026966.html</span></a>
<br>
<br><span style=" font-size:10pt;font-family:sans-serif">For each composition,
try all the other compositions as a reference composition and find the
one with the best match.</span>
<br>
<br><span style=" font-size:10pt;font-family:sans-serif">That worked reasonably
well, and found compositions such as </span><a href=https://complib.org/composition/35968><span style=" font-size:10pt;color:blue;font-family:sans-serif">https://complib.org/composition/35968</span></a><span style=" font-size:10pt;font-family:sans-serif">
which wasn't close to anything in the Full Monty list.</span>
<br><span style=" font-size:10pt;font-family:sans-serif"><br>
</span><span style=" font-size:9pt;font-family:Lucida Console">./c80532.rows=5346
rows c35968.rows=5024 rows rotate 4706:>15643278 start=16542378 same
rows=2630 52.35% same next row=2471; 93.95% of same rows; 49.18% of all
rows</span>
<br>
<br><span style=" font-size:10pt;font-family:sans-serif">As a comparison</span>
<br><a href=https://complib.org/composition/67542><span style=" font-size:10pt;color:blue;font-family:sans-serif">https://complib.org/composition/67542</span></a><span style=" font-size:10pt;font-family:sans-serif"><br>
</span><a href=https://complib.org/composition/33105><span style=" font-size:10pt;color:blue;font-family:sans-serif">https://complib.org/composition/33105</span></a><span style=" font-size:10pt;font-family:sans-serif"><br>
</span><span style=" font-size:9pt;font-family:Lucida Console">c67542.rows=5120
rows c33105.rows=5120 rows rotate 3072:>14563278 start=16523478 same
rows=5120 100.00% same next row=5112; 99.84% of same rows; 99.84% of all
rows</span><span style=" font-size:10pt;font-family:sans-serif"><br>
</span>
<br><span style=" font-size:10pt;font-family:sans-serif">look similar according
to my calculation.</span>
<br>
<br><span style=" font-size:10pt;font-family:sans-serif">These look very
similar:</span>
<br><a href=https://complib.org/composition/10029><span style=" font-size:10pt;color:blue;font-family:sans-serif">https://complib.org/composition/10029</span></a><span style=" font-size:10pt;font-family:sans-serif"><br>
</span><a href=https://complib.org/composition/33105><span style=" font-size:10pt;color:blue;font-family:sans-serif">https://complib.org/composition/67414</span></a><span style=" font-size:10pt;font-family:sans-serif"><br>
</span><span style=" font-size:9pt;font-family:Lucida Console">./c10029.rows=5600
rows c67414.rows=5184 rows rotate 0:>12345678 start=12345678 same rows=5184
100.00% same next row=5181; 99.94% of same rows; 99.94% of all rows</span>
<br>
<br><span style=" font-size:10pt;font-family:sans-serif">If the conductor
was calling #67414, but you thought the conductor was calling #10029 only
3 times would you be surprised by the next row if you were just considering
the current row.</span>
<br>
<br><span style=" font-size:10pt;font-family:sans-serif">Perhaps that calculation
should just look at the lead end / lead head pairs, or lead end / lead
head pairs at a tenors together calling position but it works okay for
this situation. Consider just the lead end / lead head might be better
for methods where a different course could be singled in at one of two
places - should every row pair in the singled in course be counted as a
difference or just the calls?</span>
<br>
<br><span style=" font-size:10pt;font-family:sans-serif">Andrew Johnson</span>
<br><span style=" font-size:10pt;font-family:sans-serif">Twyford</span>
<br>
<br>
<br>
<br>
<br>
<br>
<br><span style=" font-size:10pt;font-family:sans-serif"><br>
Unless stated otherwise above:<br>
IBM United Kingdom Limited - Registered in England and Wales with number
741598. <br>
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6
3AU<br>
</span>