Racing Victoria Limited v Riley [2016] VSCA 230

1 Horse racing in Australia is a highly regulated industry. This is hardly surprising, given the high level of public interest and participation and the consequent need to maintain public confidence in the industry. As has long been recognised, regulation is necessary both to ensure fair and open competition in racing and to maintain the health and wellbeing of horses and their jockeys.[1]

2 Like most sports, horse racing has strict prohibitions against the use of performance-enhancing substances. The Australian Rules of Racing (the 'Australian Rules') list some 80 categories of prohibited substances. Most are prohibited absolutely. In respect of some substances, however, the Australian Rules provide that they cease to be prohibited if present 'at or below' a specified concentration. The present case concerns what are called 'alkalinising agents', which are prohibited substances except when present at or below total carbon-dioxide (TCO2) at a concentration of 36.0 millimoles per litre ('mmol/L') in plasma.[2]

3 At the direction of the stewards, a blood sample was taken on 13 July 2014 from a horse trained by the respondent, Mr Riley. Testing revealed a TCO2 concentration of 37.061 (mmol/L) per litre. The testing laboratory 'rounded up' that figure to 37.1 mmol/L, for the purpose of expressing the reading to one decimal place. In the certificate which it provided to the applicant, Racing Victoria Limited ('RVL'), the laboratory stated that the concentration was 37.1 mmol/L, but noted that the 'measurement uncertainty' was 1.0 mmol/L. Allowing for the measurement uncertainty, the reading was treated as a reading of 36.1 mmol/L.

4 On the basis of the laboratory's certificate, the Racing Appeals and Discipline Board found Mr Riley guilty of administering a prohibited substance for the purpose of affecting the performance of the horse. The Victorian Civil and Administrative Tribunal affirmed that decision.[3] On appeal to the Trial Division of this Court, Bell J set aside the Tribunal's decision, upholding Mr Riley's submission that the Rules of Racing did not permit 'rounding up'.[4]

5 For reasons which follow, we respectfully disagree. In our view, the clear intention of the Rules of Racing (by which Mr Riley agreed to be bound) was that the task of determining whether prohibited substances were present would be carried out by accredited laboratories - and reported to RVL - in accordance with the scientific methods and procedures appropriate to that task.

6 The Rules of Racing specified the permissible concentration of the relevant substances as '36.0' mmol/L. The unchallenged expert evidence before the Tribunal was that when - as here - a prescribed concentration is expressed to one decimal place, rounding is the means by which an accredited laboratory expresses its measurement of concentration to one decimal place. The Tribunal was correct in so concluding.

7 Accordingly, the appeal will be allowed and the decision of the Tribunal restored.

8 The Rules of Racing promulgated by RVL (the 'Rules') are constituted by:

(a) the Australian Rules, made by the Australian Racing Board; and

(b) the Local Rules (the 'local rules') made by Racing Victoria.

The particular rules relevant to this appeal are part of the Australian Rules, as evidenced by the prefix 'AR'. The local rules are designated by the prefix 'LR'.

The Committee of any Club or the Stewards may penalise:

...

(h) Any person who administers, or causes to be administered, to a horse any prohibited substance:

(i) for the purpose of affecting the performance or behaviour of a horse in a race or of preventing its starting in a race; ...

10 As mentioned earlier, the Rules list categories of substances as prohibited substances. The relevant category for present purposes is that of 'alkalinising agents'.[5] AR 178C provides as follows:

(1) The following prohibited substances when present at or below the concentrations respectively set out are excepted from the provisions of AR 178B and AR 178H:

(a) Alkalinising agents, when evidenced by total carbon dioxide (TCO2) at a concentration of 36.0 millimoles per litre in plasma.[6]

11 AR 178D(1) requires that samples taken from horses must be analysed 'only by an Official Racing Laboratory'. The term 'Official Racing Laboratory' is defined in AR 1 to mean:

an analytical racing laboratory that is accredited by the National Association of Testing Authorities ... and is approved by the Australian Racing Board and published in the Racing Calendar.

A note to the definition identifies a number of laboratories as having been approved by the Australian Racing Board. The only one relevant for present purposes is Racing Analytical Services Limited ('RASL').

12 AR 178D contains provisions governing the analysis of blood samples taken from horses. Those relevant to the present case are rules 178D(2)-(3), which provide as follows:

(2) Upon the detection by an Official Racing Laboratory of a prohibited substance in a sample taken from a horse such laboratory shall -

(a) notify its finding to the stewards, who shall thereupon notify the trainer of the horse of such finding; and

(b) nominate another Official Racing Laboratory and refer to it the reserve portion of the same sample and, except in the case of a blood sample, the control of the same sample, together with advice as to the identity of the prohibited substance detected.

(3) In the event of the other Official Racing Laboratory detecting the same prohibited substance ... in the referred reserve portion of the sample and not in the referred portion of control the certified findings of both official racing laboratories shall be prima facie evidence that a prohibited substance has been detected in that sample for the purposes of these rules.

Scientific measurement and the expression of results

13 RVL led expert evidence in the Tribunal from Dr John Vine, who is a consultant to RASL. Dr Vine's statement, which was not challenged, explained in some detail how a laboratory obtains, from the National Association of Testing Authorities ('NATA'), the accreditation which the Rules require. He also explained how samples are analysed and concentrations of TCO2 calculated.

14 According to Dr Vine, in order to obtain NATA accreditation a laboratory must provide all of its testing documentation for review by NATA assessors. The assessment is carried out against the current version of the relevant international standard.[7] Dr Vine's statement notes that:

Each method is reviewed in terms of its scientific validity and its suitability for the analyte(s) being tested. Quantitative methods, such as RASL's TCO2 method, are specifically examined in relation to their accuracy and the proper calculation of measurement uncertainty. Laboratories are reassessed every 2-3 years and their methods and quality control results are reviewed to ensure continuing compliance and performance.[8]

15 As Osborn JA pointed out in the course of the hearing, RASL's 'Scope of Accreditation' from NATA included the following, under the heading 'other substances':

Qualitative identification of prohibited substances (and their metabolites and artefacts) as defined by National and State Racing Authority Rules

Analysis of equine plasma

Analysis by ISE techniques by the methods of

RASL in-house method for the following determinations -

Quantitative analysis of total carbon dioxide.

16 In other words, RASL had specific accreditation for its method of determining TCO2 concentrations. Applying that method in this case, RASL calculated the TCO2 as 37.061 mmol/L. According to Dr Vine, that value was 'rounded up to 37.1 mmol/L in accordance with the current Australian Standard'. Dr Vine said that rounding to one decimal place

brings the calculated value in line with the threshold of 36.0 mmol/L in the Rules of Racing which is specified to one decimal place.

17 The Australian Standard[9] contains the following Statement of Principle:

Rounding a number (sometimes called 'rounding off') consists in the rejection of any figures beyond the number of significant figures which it is desired to retain and, if necessary, in the light of the rejected figures, the adjustment of the last figure or figures retained. The unrounded value should embody all the useful information that can be extracted from the data or the computation on which it is based.[10]

18 The Standard describes the 'rounding method' as one of two 'commonly accepted methods in interpreting specified limiting values'. Under this method:

specified limiting values of 2.5 mm max, 2.50 mm max and 2.500 mm max are taken to imply that, for the purpose of determining compliance with specifications, an observed value or a calculated value should be rounded to the nearest 0.1 mm, 0.01 mm and 0.001 mm, respectively, and then compared with the specified limiting value.[11]

where it is the intent that a limited number of places of figures in an observed value or a calculated value are to be considered significant for purposes of determining compliance with a specification.[12]

19 As will appear, until 2007 RASL had 'rounded down' all measurements of TCO2 concentrations. The adoption of the rounding method described in the Australian Standard accompanied a change in the quantification of the measurement uncertainty.

20 At first instance, as on the appeal, RVL submitted that the regime for scientific testing of samples, and the specification of the relevant limit as '36.0', entailed the conclusion that the calculation of the concentration of TCO2 in a blood sample should be carried out, and expressed, in accordance with the conventional (and accredited) rules for such calculation and expression. The trial judge rejected this submission, saying:

I cannot accept this submission because the relevant clauses of the Rules are equally susceptible to an opposite interpretation. Nothing in the Rules points unmistakably and unambiguously to an interpretation that rounding-up is permitted. The court will refuse to adopt an interpretation that, in effect, enlarges the category of proscribed acts to cover Mr Riley's case when the Rules do not unmistakably and unambiguously so provide. Another reasonable interpretation is equally available. This is consistent with the interpretative principles that are applied by courts in such situations, especially the principle of legality and the principle applying to provisions or clauses that are penal in nature or operation, which I have set out.

In my view, it is necessary to appreciate and follow through the importance of the specification of the decimal place in the clauses of the Rules concerned. In the case of cl 178C(1)(a), which applies here, the specified concentration of TCO2 is 36.0 millimoles per litre, which is expressed to one decimal place. The analytical task is to determine whether TCO2 is present in the sample 'at or below' that level of concentration. If it is 'at or below' that level of concentration, the substance is not prohibited and no offence has been committed. If it is above that level of concentration, the substance is prohibited and an offence has been committed. Noting that the threshold is specified to one decimal place, a sample will contain TCO2 above 36.0 millimoles per litre if it contains TCO2 in an amount that is at or above the next highest level of concentration in the applicable decimal series. Again, noting the significance of the decimal place, the amount that is next highest to 36.0 millimoles per litre in the applicable decimal series is 36.1 millimoles per litre. It follows that, in terms of a positive result, what is analytically significant under the Rules is whether the concentration of TCO2 in the sample is 36.1 millimoles per litre or higher. Because the concentration value in cl 178C(1)(a) is specified in terms of one decimal place, results between 36.0 and (up to but not including) 36.1 millimoles per litre are negative, that is, not above 36.0 millimoles per litre.

This analysis can be illustrated by reference to the facts of the present case. The adjusted RASL result was that the sample contained TCO2 in the concentration of 36.061 millimoles per litre. Was that in excess of the threshold of 36.0 millimoles per litre specified in cl 178C(1)(a)? No it was not. The clause does not admit consideration of analytical results going beyond the first decimal place. It requires consideration of whether the threshold has been exceeded in terms of that decimal place. Properly interpreted, cl 178C(1)(a) contains a rule of decimal specificity: a horse's blood will contain TCO2 that is above the threshold concentration of 36.0 millimoles per litre only if it is present in a concentration of 36.1 millimoles per litre or greater. The adjusted RASL test result of 36.061 millimoles per litre is less than 36.1 and therefore not more than 36.0 millimoles per litre.[13]

21 On the appeal, counsel for Mr Riley sought to maintain this analysis. According to the submission, the Rules - which were said to be 'silent on the question of rounding up' - were susceptible of two competing interpretations. The first was that rounding up was permitted; the second was that it was not permitted and that what was required instead was 'rounding down' or 'truncation'. (As the trial judge explained by reference to American authority, 'truncating' in the present case would have meant reporting simply the first decimal place and omitting the second and third decimal places. Thus a reading of 0.079 would be reported as 0.07 and a reading of 0.089 as 0.08.[14])

22 In our respectful view, no question of competing interpretations arose. This was not a case where the Court was required to choose between competing interpretations of words used in a statutory or contractual provision. In the present case, the words used in the relevant Rules were clear and unambiguous.

23 Rather, what was in issue here was a numerical expression '36.0 millimoles per litre'. In the context of the Rules, that expression was used to define the task of accredited laboratories in determining whether an alkalinising agent was present at a concentration which would make it a prohibited substance.

24 As a licensed trainer, Mr Riley agreed to be bound by the Rules.[15] He was therefore bound to accept the determination of an accredited laboratory on the question of concentration, provided that it was arrived at in accordance with the methods and procedures which were properly applicable to the task.

25 Accordingly, the questions for determination by the Tribunal were as follows:

(a) what did the numerical expression '36.0 millimoles per litre' convey to the laboratory about how it should express its own determination of the TCO2 concentration in a particular sample?; and

(b) what rules governed the conversion of a raw measurement of concentration levels into the required expression of a TCO2 concentration?

26 Unsurprisingly, the Rules themselves did not provide any direction or guidance on these technical issues. They were matters of science which the laboratories would be expected to resolve. Hence the answers to both questions depended on identifying the applicable rules within the relevant scientific discipline.

27 Just as there are language conventions which inform the interpretation of words and phrases, so there are 'number conventions' which govern how numbers may be expressed for scientific (and other) purposes, and which enable an informed reader to understand what particular forms of numerical expression are intended to convey. In Smith & Nephew PLC v Convatec Technologies Inc,[16] in a passage which his Honour quoted,[17] the English Court of Appeal explained how 'number conventions' govern the meaning of numbers in a scientific context:

[I]t is standard practice for scientists deliberately to express numerical values to a particular degree of accuracy and, as a result, the degree of precision with which any particular number is written conveys to the reader how the author intended the number to be understood. This may be important for a number of reasons. It may, for example, allow the author to convey the level of accuracy with which a calculation needs to be performed or it may indicate the experimental uncertainty in a measurement which needs to be taken. In all such cases the precision with which the number is expressed is understood by the reader to be dictated by the number of significant figures or decimal places the number has. It is also generally accepted that in order to compare one number which is expressed to a particular degree of precision with another which is expressed to a different or apparently greater degree of precision then it may be necessary to round the second number to the same degree of precision as the first, and to do so in accordance with the standard rounding convention ...

These number conventions have been recognised in a series of cases in this country involving the interpretation of patent claims with numerical ranges.[18]

28 Ordinarily when a disputed question of this kind arises, a court will be reliant on expert evidence as to what the applicable number conventions are and (if relevant) how they are to be applied in the particular context. In the present case, the evidence of Dr Vine was led for that purpose. There was no suggestion that his relationship with RASL precluded him from giving expert evidence before the Tribunal. (The judge ruled part of Dr Vine's evidence inadmissible, but it was evidence directed at an alternative case advanced by RVL.)

29 Dr Vine's evidence established that three number conventions were applicable. The first was that the expression '36.0' in the Rules would be understood to mean that a laboratory's determination of the TCO2 concentration in a particular blood sample should be expressed to one decimal place.[19] (As appears from the extract from the Australian Standard, this convention is treated as part of the 'rounding method'.)[20]

30 The second applicable convention was the allowance for measurement uncertainty. Attached to Dr Vine's statement was a published article (of which he was a co-author) which dealt specifically with the measurement of concentrations of alkalinising agents. The article noted that, in 2006, 'the four Australian racing analytical laboratories' had decided to review 'the uncertainty budget' for results of this kind of analysis

in the context of data from positive and negative samples collected in recent years at Australian racecourses.

31 The authors' conclusion was that the then-current allowance for measurement uncertainty of 1.2 mmol/L 'appeared to be greater than necessary'. They recommended a reduction in the allowance to 1.0 mmol/L. The recommended measurement uncertainty was derived from calculations which took into account the possibility of a rounding error of up to 0.05 mmol/L. It was subsequently adopted by all codes of racing.

32 That change in turn affected the third applicable number convention, the rounding convention. Until then, as noted earlier, the rounding convention applied by the laboratories was that measurements should be rounded down. But once the allowance for measurement uncertainty changed, so did the rounding convention. Dr Vine explained the change in an October 2007 memorandum to staff at RASL:

Following the adoption of the new measurement uncertainty of 1.0mmol/L by all codes of racing, the policy of rounding down in confirmatory TCO2 testing is to cease.

The new calculation of measurement uncertainty includes an allowance for rounding.

Rounding is carried out to one decimal place using the normal rules for rounding (Australian Standard AS2076-1984). Some relevant examples are shown below.

37.050 rounds to 37.0

37.051 rounds to 37.1

37.349 rounds to 37.3

37.350 rounds to 37.4

37.450 rounds to 37.4

37.451 rounds to 37.5

33 The trial judge noted this change, and the circumstances in which it occurred, and said:

I will not be going into the relationship between adjustment for measurement uncertainty and rounding-up as the latter is a matter of interpretation under the Rules, not determination by policy.[21]

Likewise, the reasons given by VCAT for concluding that rounding-up 'is justified in cases of this nature' cannot be accepted. While rounding-up is generally accepted by scientists, mandated by AS2706-2003 and (may be) undertaken by official racing laboratories following NATA protocols, it is a different question whether it is permitted under the Rules. Interpretation of the Rules is not a question of fact. Further, the Rules are autonomous and not to be interpreted by reference to the policy-based adjustment for measurement uncertainty of 1 millimole per litre. When, in the interpretation and application of the Rules, due attention is paid to the importance of the specification of the threshold to one decimal place, it is not, with respect, 'absurd and unreal' to propose that '36.061 is not in excess of 36 [sic]'.[22]

34 As counsel for RVL pointed out, however, it was not correct to describe either rounding-up or measurement uncertainty as issues of 'policy'. As we have explained, the evidence simply identified the number conventions which, at the relevant time, governed the work of the accredited laboratories to whom the task of determination was committed. The adoption of the standard rounding convention was a decision of the laboratory itself, and it was an element of the testing procedures by which Mr Riley (like every other trainer) had agreed to be bound.

35 As can be seen from the reasons extracted above, his Honour concluded that a different number convention - that of truncation - was equally applicable and was to be preferred because it would produce what his Honour considered to be less harsh results in a case such as Mr Riley's. With great respect, given that the unchallenged expert evidence identified the applicable number conventions and confirmed that they had been correctly applied, no occasion arose to consider whether some other convention might have been adopted or applied.

36 As noted earlier, on Dr Vine's evidence the calculation of measurement uncertainty at 1.0 mmol/L made allowance for rounding error up to a maximum value of 0.05 mmol/L. The submission for RVL was that, as a result, any 'unfairness' which might otherwise result from rounding up was eliminated. For the reasons we have given, the only question for determination was whether the methodology adopted was authorised by the Rules. With the possible exception of demonstrated unfairness, which might have shed light on the question of contractual interpretation, no separate question of fairness arose.

37 For these reasons, leave to appeal will be granted and the appeal allowed.

[1] See, eg, R v Disciplinary Committee of Jockey Club ex parte Aga Khan [1992] EWCA Civ 7; [1993] 2 All ER 853, 857-8; Harper v Racing Appeal Tribunal (1995) 12 WAR 337, 347.

[2] Racing Australia, Australian Rules of Racing (1 July 2014) r 178C(1)(a).

[3] Riley v Racing Victoria Limited (Unreported, Victorian Civil and Administrative Tribunal, Senior Member Williams, 1 April 2015).

[4] Riley v Racing Victoria Limited [2015] VSC 527 ('Reasons').

[7] International Standard ISO/IEC 17025:2005 - General requirements for the competence of testing and calibration laboratories

[9] Australian Standard AS 2706 - 2003: Numerical values - Rounding and interpretation of limiting values.

[13] Reasons [62]-[64] (emphasis added).

[15] Australian Rules AR.2; Local Rules LR.3(1)(a).

[16] [2015] RPC 31.

[19] See [16] above.

[20] See [18] above.