[2025] UKUT 00138 (LC)

It is clear that the Tribunal does not have a complete set of accurate data about NR’s demand or requirement from 2017/18 onwards to date (which, since this decision is issued in April 2025, we take as the end of the financial year 2024/25). However, we do accept as accurate figures that are unchallenged, in particular the claimant’s sales data, NR’s withdrawals from the stockpile from 2019/20 onwards, and NR’s requirement from 2020/21 onwards.

We also accept that 4,000 sleepers were withdrawn from the stockpile in 2018/19, in the absence of anything specific from Mr Jarvis.

So the real challenge is to the years 2017/18 to 2019/20. Mr Jarvis’s recollection was that NR’s requirement was close to 500,000 in each of those years. NR says it purchased 327,796, 369,390 and 292,853 sleepers respectively; when added to the withdrawals from the stockpile the total requirement was:

We know that there is an internal inconsistency within NR’s own data for those years; as the claimant demonstrated, the figures given for the claimant’s sales to track and for its contributions to the stockpile cannot both be right. The figures given for the claimant’s sales to track (table 1a) are a component of NR’s figures for its total requirement (table 1), and in light of Mr Jarvis’s recollection we agree with him that the figures given by NR for its requirement for these three years must be too low.

We said above that we do not know how NR’s figures were produced; but there is an exception to that. We have been told the outcome of a physical count of the stockpile in 2020, and of another in July 2022. Mr Heubeck observed that the count in July 2022, when combined with what was withdrawn by that date, indicates a stockpile of around 820,000 sleepers (we round the number because the figure for withdrawals between April and July is an estimate). We know that at some point in 2020 the stockpile stood at around 725,000 (paragraph 59 above), and we have accepted that by 31 March 2020 about 75,000 sleepers had been withdrawn. We do not know when in 2020 the count was made, but if, as in 2022, the count took place in the summer, then more than 75,000 had probably been withdrawn by then, again consistent with a stockpile of something over 800,000 sleepers. We are well aware that the count might have been made in December, in which case more would have been withdrawn and the total would be higher. But we think that it is significant that if we take the sort of quantity likely to have been withdrawn by June or July we get a figure consistent with the 2022 count. That may be a coincidence, but we take the view that it is more likely to mean that the physical stock counts were not so unreliable after all. We find, on the balance of probabilities, that the stockpile comprised, very roughly, in the region of 820,000 sleepers.

That indicates, as the claimant suggested, that Mr Heubeck’s figure for what the claimant sent to the stockpile was too high.

It also means that NR’s figures for the contributions made by TWM and the claimant (320,000 and 300,000 respectively, see paragraph 74 above) are too low. If we add to those two figures the contribution of 50,000 from Tallington, we have about 670,000, which is about 150,000 too low. There is no challenge to the figure from Tallington and we conclude that TWM and the claimant together sent 770,000 sleepers to the stockpile, so that the figures given by NR for the two companies’ contributions are, taken together, wrong (as NR itself indicated by adding a note to the effect that the physical count of the stockpile showed “~100,000” more) by about 150,000. The error may be in the TWM figure, the claimant’s figure, or both. We have no information about TWM’s sales.

Accordingly we have to conclude that the claimant sent no less than 300,000 and no more than 450,000 to the stockpile. Any less than 300,000, and the stockpile is too small. Any more than 450,000, and it is too big (as a little experimentation with table 2 above demonstrates). If the claimant’s contribution to the stockpile was 450,000 (so that all the error is in NR’s figure of 300,000 for the claimant’s contribution) then NR’s figures for what the claimant sent to track are 75,000 too low (again, this can be seen by playing with table 2). If the claimant’s contribution was 300,000 then NR’s figures for what the claimant sent to track are 225,000 too low.

We have no way of determining how big is the error in what NR says the claimant sent to track. But we do not have to decide how many sleepers the claimant sent to the stockpile; we have to decide NR’s overall requirement, which will determine what the claimant would have sold in the no scheme world once we have determined its market share in that world (issue 3 below). The figures given by NR for what the claimant sold for work on the track in these three years can be seen to have been between 75,000 and 225,000 too low. We have no basis however for saying that what NR said TWM sold to the track (again, components of the figures in column 2 of table 1, which we have not set out here) was incorrect.

NR’s own figures give its total requirement in these three years as 1,065,636 (see the final column of table 1 above, and paragraph 77). Mr Jarvis’s recollection that the figures were nearer 500,000 in each of those three years indicates that NR’s totals are over 400,000 short. We bear in mind that Mr Jarvis does not claim to recall precise figures, but we can rely on his memory at least so far as to say the error must be at the upper end of the range of 75,000 to 225,000. We find therefore (a) that NR’s figures for what it required for track in total in the three years 2017/18 to 2019/20 were too low and (b) that they were too low by about 225,000. It is fair to say that we have far more confidence in (a) than (b), but we nevertheless make that second finding on the balance of probabilities.

We are conscious that we have laid a lot of weight on the size of the stockpile, on physical counting of the stockpile, and on the figures for withdrawals from the stockpile. As we say, we are doing the best we can on the basis of the limited and conflicting information we have. We make the following findings of fact:

The accountants need figures to work on for the years from 2017/18 to date. If we share the 225,000 shortfall arbitrarily between those three years, purely for the sake of the accountants’ calculations, then NR’s total requirement to date (giving rounded figures for the years where we have added an approximate figure) was:

We do not comment in this interim decision on the accountants’ methodology, but the forensic accounting experts should use these figures for NR’s requirement to date in the no scheme world rather than constructing their own version of that requirement by re-distributing, in the no scheme world, the sleepers sold for the stockpile in the real world.