Since 1991 Core laboratories have been testing growers’ cane for determining payment at Jamaican factories. Now the testers have become the tested.
Cane growers anywhere ought to feel confident that, regardless of which core lab does the test, results will be similar, allowing for the natural variation that occurs within the sample itself. Thus growers in Trelawny, for instance, should not have to contemplate whether results may be different whether cane is tested at Long Pond or at Hampden. To guard against this possibility, SIRI has been conducting a series of tests to see that all core labs function as they should.
Core labs across the island are all fitted with the same type of equipment and the same testing procedure should always be followed. However, with use, all instruments will eventually need readjustment. Continuous monitoring is therefore necessary to ensure that all things are in order, and if they are not, to fix them quickly.
The best way of determining accuracy is to compare results obtained at one core lab with those obtained at others working on the same material under similar conditions.
In order to do this, painstaking efforts are made to select mature healthy, damage-free stalks of one variety, of same age, grown under the same conditions in a small section of a field. All stalks are similarly cut at the base, topped, stripped of trash and cleaned of adhering dirt. This forms the major sample which will be distributed between labs.
Even with all this careful selection and preparation, it is known that there are large differences between cane stalks. So, each stalk is then cut into billets (joints) to allow each sub-sample to contain a piece of each cane. In the end, each sub-sample has an equal number of bottom middle and top joints. Sub-samples are then sent under similar conditions of storage and transport to the various labs with instructions for the identical set of analyses to be conducted at the same date and time.
Data from these analyses are then compared. Ideally, every lab should give exactly the same result. But this is never so in practice. Allowance must always be made for a certain level of “error.” So long as this is kept within certain limits, then the results are acceptable. In these tests acceptability is usually measured statistically by a “standard deviation.” An average value is first established and then the variance from that average calculated. From this the standard deviation is determined. As a rule, the smaller the standard deviation the better. This would indicate that the labs being compared are giving more or less the same result. The question then is what is a large or small standard deviation in cane test results?
During the Mills Commission of Enquiry from which the present Jamaican cane payment system was recommended, one of the main reference points was the work of Dr. Saranin from Australia. He carried out a number of tests on cane samples from which we could make useful comparisons. Conducting 16 determinations of % pol on green cane, which would be the closest equivalent to the SIRI tests, he found a standard deviation of 2.21. Standard deviations on pol in the SIRI tests have all fallen below this level, Tables 1-4, and should therefore tell us that our performance is at least comparable.
Another measure of a lab’s performance is its precision, that is, how repeatable the results are within the same lab. For these tests, within each lab the same analyses are run on each sample six times and results compared. Again, the smaller the variation, the better the lab’s performance. To find out how precisely the lab determines the true value another statistical test called a coefficient of variation (CV) is calculated. The smaller the CV, the greater is the lab’s precision. In field experiments a CV of up to 25% is considered acceptable. For laboratory analyses a much smaller CV would be desirable. This CV is applied to both within lab and between lab comparisons.
Tests include analyses of juice brix, pol and purity, bagasse quality and quantity, cane pol and % fibre, all towards determining the Jamaican recoverable cane sugar, or JRCS. In the tests within the labs, it is usual to find greater precision in determining juice than fibre data. Thus the CV for brix, pol, purity ranges between 0.77 and 1.07. But fibre analyses show a CV of 15.82% within the lab, Table 1. Where a number of labs are compared the CV falls ( to 5-12%), Tables 2-4, as the sample size increases by the number of participating labs and replications within the labs.
The test of particular interest to the grower is of course the JRCS, as this determines cane payment. Percent pol in juice and fibre in cane are the main components used in determining the JRCS. As mentioned earlier, CV for pol % juice seldom exceeds 1% when a comparison is done within the laboratory. However, between laboratories CV values as high as 4.86% (which are still quite acceptable) are occasionally observed, Table 2. Results from individual labs are encoded A-H in the tables.
Variations in JRCS readings more frequently result from less precise fibre related determinations. In that example of repeated tests within a particular lab, Table 1, CV for bagasse % cane was 1.5%, fibre % cane -15.8% and bagasse % moisture - 10.8%. A more exact determination of the fibre would give a more reliable JRCS reading; hence the need to accurately weigh and dry the bagasse during fibre analyses.
Unlike juice analyses therefore, fibre determinations tend to be quite variable, largely because of the relative difficulty in obtaining a truly homogenous fibre sample, which in turn is a result of the highly variable nature of cane even along a single stalk. Over the years 1996-98, labs showed CV values ranging between roughly 6 and 15% in determining cane fibre, Tables 2-4.
A major concern is in the measurement of moisture in bagasse. Extreme values are often observed. Unusually high moisture levels will result in low fibre % cane values and vice-versa. The variations suggest that samples are not always dried to constant weight, or, sample weights may not be accurately taken.
Given the natural variability within cane samples, the tests show the core labs to be maintaining acceptable levels of accuracy and precision in cane testing. The industry may therefore be reassured that testing gives a fair approximation of cane quality.
Growers will always find wild fluctuations in tests between one cart and the next because, unlike these tests in which samples were carefully selected, in testing the typical cane truck or cart, the passage of the core tube through the load is unpredictable and composition of the sample highly variable. Differences, and often large differences, are therefore to be expected and will occur because of the inherent nature of cane and the relatively crude process of harvesting. The sucrose (pol) in juice will vary along the length of the stalk, and chances are that suckers, tops, roots, soil and debris which are often incorporated in loads might also be cored.
Special attention is being given to accuracy in weighing samples and drying to constant weight for moisture determinations. This could account for the lowered CV (6.1%) for cane fibre analyses in the last series of tests, Table 4.
Ongoing training along with close monitoring to ensure that established procedures are faithfully followed should result in the core labs continuing to pass the test.