Problems with Ball Mass and Bounce Testing
Extract from letter to President of SA Croquet Association - 28 September 1989.
Thank you for your copies of correspondence between yourself and Mr C. Jaques regarding the 23 balls which Messrs John Jaques have so kindly supplied in place of the 13 sent. A very generous reply to your complaint.
1. Now, you have asked for further comment regarding the disparity in results of tests obtained here and in the U.K. on the 13 balls in dispute. Please refer to the enclosed chart, of which I have kept a copy, of some of the results obtained on the balls in question.
1.1 Unfortunately, mass results were not recorded for any balls apart from unplayed new balls so marked with red ticks on the chart. The reason was that old balls cannot really be rejected for mass even if they have only been used very few times. The same goes for size and bounce, legally speaking, except that the bounce normally increases with flattening of the milling after use. This is one reason which makes the Barlow balls attractive, in that they hardly seem to wear, even after prolonged use.
1.2 Regarding mass it will be seen from the chart of results that the mass of the white ball of set 7 was recorded as 468.1 grams, which is presumably to be compared with the mass of 466 grams on Mr Jaques' list. Similarly, the yellow ball of set 8 was recorded as 435.7 grams, to be compared with either the 434 or 436 grams of Mr Jaques.
1.3 Since the mass of the white ball was found by Mr Jaques to be 2.1 grams lighter than the mass recorded on the chart, it is presumed that the yellow ball corresponding to this had a mass of 434 grams according to Mr Jaques. some 1.7 grams lighter. The discrepancy between 2.1 and 1.7 is accounted for by rounding the figures off to the nearest whole number in each case.
1.4 After the above comments on mass, I would like to point out that the chart made mention of only those balls with a green tick which could legitimately be returned. These comprised three balls, namely the blue and red of set 14, and the pink of set 6. I was accordingly a little surprised to see that 13 balls had been returned, albeit that all 13 were outside specification.
2. This leads me to the vexed question of bounce. The problems here are non homogeneity, and/or temperature. To explain:-
2.1 Firstly, bounce can vary from one spot to another on any one ball. For this reason tests should be conducted on known points if comparisons are to be made between different test apparatus. In order to do this, the resilience apparatus must be capable of bouncing a ball on or close to any particular spot. I normally use the four poles and two nodes of each ball, three times on each point, or 18 bounces in all per ball. The procedure is only reduced if the balls of any one type prove to be monotonously accurate.
2.2 Secondly, resilience changes with temperature. Up to 20% over 10° C on nylon, less with other plastics, and controllably less with special compounding techniques involving additives which can stabilise these effects.
Normally a ball should be tested at ambient conditions, being 1 bar, 55% relative humidity and about 20° C. These are not absolute figures but represent conditions of climatic comfort obtaining in most parts of the populated world at some time of the year. With any fresh unplayed set of new balls of any make, where the results must be guaranteed, I soak the balls to be tested in a bowl of water, kept at 20° C either by the addition of ice blocks or hot water, for about one hour. Then they are tested with the customary 18 bounces.
2.3 In the case of the 13 returned balls, however, I must admit that I did not have the time to do the normal water soak, and carried out the test late one afternoon when the ambient temperature had fallen to about 70° F or so. It is for this reason that I suggested on my results sheet that only the green tick marked balls could be legitimately rejected. It is quite possible that the balls were somewhat warmer than 70° F which would account for up to l" variation with the Jaques composition, but when the results that Chris Jaques obtained are examined, there is obviously some other factor present to account for the differences between those results and my own.
2.4 For obvious reasons, it has not been possible to correlate the two sets of figures precisely, since I did not keep a record -of the individual single colours by, for instance mass, which Chris Jaques has included in his list.
However, since there was only one red, one pink and one green, these can be correlated. Also the two whites, since on my sheet I marked the overmass ball of set 7 with a brown cross, while recording its bounce at 31". This ball must correspond as the white on the Jaque's list of mass 366 grams, which also leaves the other white identified. The following list is therefore a reasonable comparison between the two sets of bounce figures.
Attention is drawn to the difference between the two test results, tabulated so as to reduce these to a minimum (e.g. the blue balls show a maximum possible variation of 1.75 against the 1.25 inches shown in the difference column).
These differences cannot be understood by temperature variation alone, even if the U.K. tests were undertaken in widely differing weather conditions. Supposing that the U.K. tests were undertaken at a reasonably constant temperature, then my neglect to soak the balls for one hour at 68° F could account for the single negative difference of -0.25" and the two cases of +0.25". Even the 3 cases of 1.25" difference would be acceptable at a pinch, but then we diverge through 1.5", 2", 3.25", 3.5" up to 3.75". This is not acceptable, and I have decided to try to settle these extraordinary discrepancies by a further test.
2.6 For the test I have taken one of my own old Jaques balls, marked only 'composition' on one pole and "A" on the opposite pole. The precise age and quality of the ball is completely irrelevant to the test, and in fact I think the problem of homogeneity can be best elucidated in the tests on this ball. I have used the ball for garden practice for years.
2.7 Before you handle it, I caution you to handle it with extreme care, if at all. The ball was marked by me with a black fibre pen. One pole with the Jaques legend "COMPOSITE" on it was left unmarked, as was the pole marked "A". One of the other poles was marked with a bar like the figure 1. and the opposite pole with a cross. Lastly, one of the nodes was marked with a black circle. and the other node with a black square.
2.8 The reason that I ask you to handle the ball with care is to avoid rubbing the faint marks of white paint off the various black markings. These white paint marks come from the bouncing steel plate, which although now reasonably bald over about 10 mm diameter, still marks the balls a little. This plate is only painted infrequently, and very thinly, but there was enough left on to just make faint marks. These marks demonstrate clearly the accuracy with which the drop can be executed to impact any chosen spot within 5mm plus-minus. As you will see from the bounce results, this is important for homogeneity tests as you also saw when you visited the test site some time ago.
2.9 Meanwhile, concerning the requirements of solidity and hence resilience of the test rig, I am certain that the mass of high strength concrete and the high strength bonding of the thoroughly rust treated and embedded steel plate are more than adequate, and certainly exceed the legal requirements according to the rules. The mass of concrete immediately surrounding the plate is 300 mm thick by 500 mm in diameter, of good quality, strengthened by an admixture of one of the extra hard cements, while the plate has been embedded without air entrapment in 100% extra hard cement. The unit occupies a corner between a brick wall and a 100mm concrete floor.
3. Here follow the test results of this ball, which is a few years old now although little used as you can see.
3.1 Mass on one scale, the one I normally use 438.7 gm. On another scale, measuring just as accurately but of a different make, 438.5 gm. Size 91.45 mm across from 'composite' to "A" poles. 91.55 mm across the other poles and 91.5 mm across the nodes.
3.2 Bounce figures follow below
4.1 The fact that the separately checked mass results concur within 0.2 gms, leads me to believe that there is a possibility of a ± 2 gm error either in both the local electronic scales or the U.K. scale.
4.2 Regarding size and shape, the sphericity of the tested ball is near perfect within 0.1 mm, the size lies perfectly in the specification bracket of 91.2 - 92.8 mm, towards the lower limit which is accounted for by wear.
4.3 The bounce results are, as you can see, very close to each other as regards each group of 3, but widely disparate group from group in comparison with each other. The lowest bounce average is 25.5", while the highest is 32.25", a difference of 6.75" or 20%. The overall average is 29.l" or only just below specification.
These results on bounce highlight and clarify my call for testing on known spots, and the reason I have cautioned careful handling of the ball is that I think it should be forwarded to Chris Jaques himself to see, and for him to compare on his own apparatus precisely to achieve some kind of consensus on test results.
4.4 This concludes the reply which you requested, and whereas it cannot really be of much use retroactively, I hope that it will pave the way to more understanding between the various members of the Croquet Fraternity in the future
R. le Maître
All rights reserved © 1989