Prosecutions for water pollution offences commonly depend on evidence derived from samples. Prior to the 1989 Act, procedures were governed by section 113 of the Water Resources Act 1963, which aimed to protect the defendant by providing that on taking a sample of effluent the occupier of the land concerned must be notified, and that the sample must then be divided into three parts and sealed in containers. One container was to be given to the occupier, one taken for analysis by the authority, and one retained by the authority for future comparison. Any analysis based on the sample could not be used in subsequent legal proceedings if this "tripartite" procedure had not been followed.
These provisions were replaced by section 148 of the 1989 Act, and some commentators had assumed that no change in the law had occurred. There was, however, an important distinction in the wording. The provisions of the 1963 Act had related specifically to samples of "effluent". The drafting in the 1989 Act, however, provided that "the result of the analysis of any sample taken on behalf of the Authority in exercise of any power conferred by this Act shall not be admissible in any legal proceedings in respect of any effluent" unless the tripartite procedure had been followed.
In the Harcros case, the NRA had wished to rely on the analysis of a sample, not of the effluent itself, but of the river affected by the discharge. The tripartite procedure had not been followed in respect of that sample, but the NRA argued that the provisions were clearly based on the 1963 Act and were intended to cover effluent samples only.
The defendants pointed to the actual wording of section 148. While the proceedings themselves related to an effluent discharge, the section applied to the analysis of "any" sample taken by the NRA. This would include the sample of the river.
The Divisional Court favoured the defendant's arguments. There was no presumption against Parliament changing the law, even if that added to the burden of the NRA. "The legislative purpose was to protect the defendant in a case of impending legal proceedings," the Court concluded. "There was no distinction between a sample of effluent and a sample of river if both specimens were required for a prosecution."
The decision is consistent with the wording on the face of the legislation. The tripartite procedure, however, applies only to samples taken by or on behalf of the NRA where analyses are to be used in legal proceedings. Samples taken for monitoring purposes are not covered, and the 1989 regulations dealing with public registers of water and effluent data expressly provided that register entries concerning samples must indicate whether the tripartite procedure had been followed.
There may be situations where the NRA is not able to notify the occupier "on taking the sample" - the occupier may not be present at the time, for example, or, in the case of river water samples, may well not be identifiable immediately - but the 1989 Act specifically allowed for this by providing that if it is not "reasonably practicable" to comply with the requirements when the sample is taken, the provisions would still be satisfied if they were complied with as soon thereafter as "reasonably practicable". No case-law yet exists on the meaning of these familiar terms in this context.
Not all pollution prosecutions depend on sample analysis, and previous case-law has held that visual evidence of dead fish, for example, is admissible without the need for tripartite samples. The decision in the Harcros case does not affect that principle.
Finally, the provisions apply only where samples are taken by or on behalf of the NRA itself. Private prosecutions for water pollution offences are not restricted, and there is no need for bodies such as the Anglers Cooperative Association or Greenpeace to follow the tripartite procedure in respect of samples used in such prosecutions. Any analysis, though, based on a sample that has been subject to the tripartite split is clearly more vulnerable to challenge.