Home Page Back to list
Sound re-created: WAH 8/5/04
Prior to the casting and installation of the current peal of twelve (originally a chime of 14), Coventry had a peal of ten which were by reputation one of the finest tens of their age - they were originally cast by Pack and Chapman in 1774 (though the 6th was recast by them in 1799, and the tenor by Briant in 1804 after it was cracked in 1802). When the old bells were taken out for recasting by Gillett and Johnston in 1926, they were the subject of a famous court case. The long and interesting history of the bells at Coventry is covered in Chris Pickford's 'The Steeple, Bells, and Ringers of Coventry Cathedral', privately published in 1987, and the court case is covered in 'Coventry Bells and how they were lost', E. Alexander Young, 1928. An account of the court proceedings - a most interesting document - appears elsewhere on this site. At the time of the case, the bells had not been rung since 1885 because of structural problems with the tower, and the sound of them in changes was already a distant memory.
Young's book includes a copy of a report on the old bells by A. A. Hughes of the Whitechapel Bellfoundry (transcribed below), which gives the nominals only of the bells, together with some comments on their tuning. Chris Pickford recently suggested that the Gillett and Johnston tuning books were in Croydon library. Dickon Love kindly went to look, and discovered a full set of tuning figures for the bells, taken on 6 May 1926, just before the bells were broken up in the foundry. The availability of these figures suggested the intriguing prospect of re-creating the sound of the bells. At St. Peter Mancroft in Norwich there exists a peal of 12, cast by Pack and Chapman in 1775, apart from the tenor which has been recast. Research by David Bryant has proved that the original bells have never been retuned. As I had a set of recordings of Mancroft, taken earlier this year, creating a simulation of the Coventry bells proved straightforward.
This link will take you straight to the simulation.
Here are the tuning figures for the bells taken by G&J in 1926, together with the nominals measured by A. A. Hughes in December 1925 (the latter figures are twice the 'tap tones' measured by AAH):
(For all partials except the nominal, the intervals in cents are from the nominal of that bell. Cents of the nominals are relative to the tenor.
By adjusting the recordings of bells 2 to 11 from Mancroft, I produced a set of recordings with the same nominals as the old Coventry bells. The tuning figures of these recordings are as follows:
Here are comments on various aspects of the tuning of the bells:
All the bells have old-style seventh hums, of course. The old Coventry bells have noticeably better hums (i.e. nearer the -2400 cents octave) than the Mancroft simulations. In the back bells, this does not much matter, because as explained elsewhere it is hard for the ear to be certain about hum tuning where the hum is low in frequency. The trebles of the old Coventry peal would have sounded a little better than this simulation because of the better hums.
The primes of both sets of bells are fairly poor; moderately close to the -1200 cents octave in the back bells, pretty flat in the front ones, giving the smaller bells the classic old-style clonkety tone. The prime of the 'Mancroft' 8th is significantly sharp, a fault not present in any of the old Coventry bells.
The tierces are very comparable between the peals; a little scattered, because they were controlled by profile only, but not bad considering. They are all Just or somewhat sharp of that tuning, rather than equal tempered as would be found in many modern bells.
All the quints are quite sharp compared with modern practice, though the Coventry bells are somewhat better. As usual, quint tuning probably does not have much effect on the sound of these bells.
For the nominals, of course there is the issue of which temperament they were tuned in. In the table below are the tuning figures for the Coventry 10, compared with Just tuning. Figures are in cents:
These figures suggest Just as a good fit. Bert Hughes, in his report copied below, judged the bells against just tuning, and his notes on the tuning discrepancies correspond almost exactly with the above differences (e.g. for the 9th, 0.15 of a whole tone = 30 / 100 of a semitone). Against Just tuning, the ninth is flat, and the treble a little flat. Apart from those discrepancies, the nominal tuning is good, quite up to modern standards of +-10 cents. There is no stretch in the trebles, rather the opposite; in any case, it is not needed, as shown by the frequencies of the higher partials. In many or most 19th or 20th century peals, the tenors would have much sharper upper partials than the trebles. Here, the profiles used for the bells give a good consistency to the upper partials across the whole peal (reminiscent of some of the very early Taylor true-harmonic peals). I make here the assumption that the Coventry bells had similar upper partials to the Mancroft bells - a good assumption, given the consistency of the partials in the Mancroft bells.
The final point to make about these bells, especially the trebles, is that though they are not true-harmonic and are rather deficient by modern standards, they have a clearly delineated set of upper partials which gives them a very audible strike note that sings out above the lower partials. The plot below shows the spectrum of the simulated treble (i.e. Mancroft 2nd) with the strike partials marked. Remember that the upper strike partials are much more audible than would appear from this plot because of the preference given them by the ear.
The individual recordings of the Mancroft bells on which this simulation is based were taken on 9th February 2004, using an AKG C1000S microphone hung a few feet above the bells, and a Toshiba laptop. Recordings of bells 2 to 11 were stretched or shrunk with Wavanal to exactly match the Coventry nominals as measured by G&J - all other partials move in proportion to the nominal, so the inner tuning of each bell is preserved. The recordings were also edited with Cool Edit to be of roughly equal loudness, and to have some minor mains hum removed. The Rounds program was then used to create the simulations of change-ringing, at a peal speed of 3h 30m. Finally, the simulations were bandwidth limited with Cool Edit and converted to mp3 files.
Here, after all the introduction, are the Coventry bells, or as close to them as we may get:
Coventry - ten seconds (69kb)
Coventry - two minutes (821kb)
The above simulated recordings are as if the microphone was just above the bells, giving no allowance for the tower acoustics, which by repute are very good. Here is a simulation with a little tower acoustic added:
Coventry - with tower acoustic (821kb)
Copy of report in letter dated 11th December, 1925
Mears and Stainbank
COVENTRY CATHEDRAL BELLS
With reference to my recent examination of the bells in the Cathedral Tower, I have to report that each bell is of good, bold and pleasing tone, and they really constitute a very fine peal of ten. I give below, a table shewing, in vibrations per second, the "tap" or main note of each bell, with its degree of inaccuracy of "tune" taking the Tenor or largest as the basis.
vibs. per sec.
|Tenor||56 3/4"||31.96||278 1/2|| |
|9th||51 1/4"||23.06||308||5 1/4 vibs. per sec. flat (.15 of a note)|
|7th||43 3/8"||14.12||370||1 1/3 vibs. per sec. flat (practically correct)|
|5th||37 1/8"||9.78||463||1 vib. per sec. flat (practically correct)|
|4th||35"||9.06||521||1 1/4 vibs. per sec. flat (practically correct)|
|3rd||33 1/4"||8.41||560||3 vibs. per sec. sharp (.04 of a note)|
|2nd||31 1/2"||7.12||623||3 3/4 vibs. per sec. flat (.05 of a note)|
|Treble||30 1/8"||6.79||690||6 vibs. per sec. flat (0.08 of a note)|
The above records were taken with specially tested tuning forks.
These bells were tuned by hand, before the days of tuning machines, and forks to test the frequency of vibrations, and it is really remarkable that such a close result should have been obtained. With the exception of the 9th and treble, the faults are negligible. I do not remember ever having found an old ring of bells (especially ten) so nearly correct.
As already stated the above records refer to the "tap" or main notes. The chief harmonic tones, viz. the "hum" notes are not, however, so correct. These hum notes vary from a seventh, to an augmented seventh interval, below the tap note. The bells were of course not designed to produce an octave hum note; they were cast for ringing and the existing hum notes were intentional. It would, however, have been better had the hum notes all been in tune together as augmented sevenths. This result could now be obtained without difficulty by means of a modern tuning machine, but it is quite a matter of opinion as to whether such an alteration should now be made.
I have found that the percentage of the public (even amongst Ringers) who are able to hear the harmonic tones in bells, is very small indeed, and I know of many rings which are generally acknowledged to be very fine and even unexcelled (as indeed they are) but which I happen to know have faults both in the tap notes and harmonic tones.
The best result is not being obtained from your bells at present, for the following reasons:-
Some of the carillon hammers are resting in contact with the bells, due to fatigued springs.
The striking surface of the hammers is worn.
The hand chiming hammers, which strike the inner surface of the bells, are worn, and some of them are not striking on exactly the correct part of the sound-bow.
The remedying of these defects would make a noticeable improvement in the general effect.
I have always heard very glowing accounts of your bells, but notwithstanding this, I was not prepared to form quite so good an opinion as I have done now that I have had an opportunity of trying them.
I consider they fully justify the good reputation they have so long enjoyed, and it is my firm opinion that there is no need for any re-casting.
(signed) A. A. Hughes
Last updated May 23, 2004. Site created by Bill Hibbert, Great Bookham, Surrey