Quinn v. Leathem,  A.C. 495 at 506 (H.L.)
[A] case is only an authority for what it actually decides. I entirely deny that it can be quoted for a proposition that may seem to follow logically from it. Such a mode of reasoning assumes that the law is necessarily a logical code, whereas every lawyer must acknowledge that the law is not always logical at all.
B.S.A. Investors Ltd. v. DSB, 2007 BCCA 94, 291 DLR (4th) 186,  9 WWR 193, 69 BCLR (4th) 242; leave to appeal denied 2008 CanLII 2275 (SCC)
From the BCCA reasons
 All three judges in B.M. [B.M. v. British Columbia (Attorney General), 2004 BCCA 402] also adopted Lambert J.A.’s formulation of the McGhee [McGhee v. National Coal Board,  UKHL 11] principle in Haag [Haag v. Marshall (1989), 61 DLR (4th) 371, 1989 CanLII 236], a formulation that sets out specific requirements that had to be met in order for an inference of causation to be properly drawn. In that case, Lambert J.A. wrote at 213:
The “inference” principle derived from McGhee … is this: Where a breach of duty has occurred, and damage is shown to have arisen within the area of risk which brought the duty into being, and where the breach of duty materially increased the risk that damage of that type would occur, and where it is impossible, in a practical sense, for either party to lead evidence which would establish either that the breach of duty caused the loss or that it did not, then it is permissible to infer, as a matter of legal, though not necessarily logical, inference, that the material increase in risk arising from the breach of duty constituted a material contributing cause of the loss and as such a foundation for a finding of liability.
. . .
 In B.M., as in this case, the question was how one specific person would have reacted had the duty of care been met; as in this case, that person did not testify. (In B.M., he was dead.) Nevertheless, the majority held that there was enough evidence on the record about his character, led through other witnesses, to come to a conclusion on causation. None of this was direct evidence, of course, but as stated by Smith J.A. (at para. 168) causation “is always a matter of inference from probative facts established by the evidence.”
 I would add that there is an important difference between drawing an inference as to causation from circumstantial evidence, which is often done, and drawing an inference as to causation from no relevant evidence at all, which may be done only in the rare circumstances set out above. This is the difference alluded to by Lambert J.A. in Haag when he distinguished between a logical inference and a legal one; the legal reference should not be resorted to unless the logical inference is impossible to establish with either direct or circumstantial evidence.
[Emphasis added by me: DC.]
For some context
 Since these reasons were written, the Supreme Court of Canada has released its decision in Resurfice Corp. v. Hanke, 2007 SCC No. 7. …
 A second situation requiring an exception to the “but for” test may be where it is impossible to prove what a particular person in the causal chain would have done had the defendant not committed a negligent act or omission, thus breaking the “but for” chain of causation. … Once again, the impossibility of establishing causation and the element of injury-related risk created by the defendant are central.
 Returning then to the facts of this case, it is the second situation that is at issue in this appeal. Even though Mr. Mosly did not testify, there was still a great deal of evidence led through other witnesses that reflected upon his character. This was circumstantial evidence from which the trial judge could have drawn logical inferences as to Mr. Mosly’s likely course of action had he been confronted with a request for a second signature. It was not impossible. Furthermore, even if this evidence did not happen to be on the record, it was possible for the plaintiffs to lead it. The plaintiffs were ultimately responsible for proving that the defendant’s breach of duty made a difference.
 The question might therefore be asked: was it really a part of the plaintiffs’ burden of proof to show that had Mr. Lester asked for the second signature it would have put a stop to Mr. Mosly’s fraud. In my view, on these facts, the answer is yes. Obviously, the plaintiffs were not required to lead evidence negating every possible hypothetical situation. But here there was a sizeable body of evidence going to Mr. Mosly’s dishonesty, which was accepted by the trial judge and summarized at paras. 57 to 72 of his reasons. Consequently, the plaintiffs’ burden as to causation included the requirement to show, on a balance of probabilities, that had Mr. Lester performed his duty Mr. Mosly would have been stopped.
As an aside:
Imagine the normal Canadian practitioner-lawyer, in a common law jurisdiction – before February 7, 2007, who saw no particular reason to keep track of the doings of the U.K. House of Lords. One morning, soon after February 7, 2007, he or she (somehow) finds out about Resurfice Corp. v. Hanke, 2007 SCC 7,  1 SCR 333 and (eventually) reads about the new version of a material contribution doctrine explained in . Those lawyers east of the B.C. – Alberta border might have said “where did that come from?” The British Columbia lawyer, however, might have nodded sagely and said “Haag”.
If you care to, compare the formulation of the requirements for material contribution in Resurfice with what Haag called the McGhee inference principle. (Or, if you’re lazy, look here, starting at p. 105 where I follow the bread crumbs.
Of course, there’s the small problem of precedent in treating Haag as a source of the material contribution principle (so long as we accept that the Supreme Court of Canada has the last say on the meaning of lower court decisions, even if the Supreme Court’s comment is obiter.) Haag was referred to in Snell v. Farrell,  2 SCR 311 at 326, (1990) 72 D.L.R. (4th) 289 at 298, 1990 CanLII 70 (SCC), as case in which McGhee was applied for the reverse-0nus principle.
Decisions in Canada after Wilsher accept its interpretation of McGhee. In the circumstances in which McGhee had been previously interpreted to support a reversal of the burden of proof, an inference was now permissible to find causation, notwithstanding that causation was not proved by positive evidence: see … and Haag v. Marshall,  1 W.W.R. 361 (B.C.C.A.).
On the other hand, it’s at least plausible that, according to the BCCA in B.S.A., Snell approved the BCCA’s decision in Haag. B.S.A. states
 The trial judge turned to the principles in Snell v. Farrell,  2 S.C.R. 311 and Haag v. Marshall (1989), 39 B.C.L.R. (2d) 205 (C.A.) for assistance. After quoting a passage from Snell v. Farrell, the trial judge noted that a passage from this Court’s decision in Haag v. Marshall had been cited with approval in Snell. The trial judge quoted from p. 213 of Haag:
 In Snell at p. 326, Sopinka J. cited with approval an earlier decision of our Court of Appeal, Haag v. Marshall (1989), 39 B.C.L.R. (2d) 205, in which Lambert J.A., speaking for the majority, wrote at p. 213:
The “inference” principle derived from McGhee, and from the three Canadian cases to which I have referred, is this: Where a breach of duty has occurred, and damage is shown to have arisen within the area of risk which brought the duty into being, and where the breach of duty materially increased the risk that damage of that type would occur, and where it is impossible, in a practical sense for either party to lead evidence which would establish either that the breach of duty caused the loss or that it did not, then it is permissible to infer, as a matter of legal, though not necessarily logical, inference, that the material increase in risk arising from the breach of duty constituted a material contributing cause of the loss and as such a foundation for a finding of liability.
The legal inference permitted by the principle may be prodded along by the concept that as between an innocent plaintiff and a defendant who has committed a breach of duty to the plaintiff and by so doing materially increased the risk of loss to the plaintiff, in a situation where it is impossible, as a practical matter, to prove whether the breach of duty caused the loss, it is more in keeping with a common sense approach to causation as a tool of justice, to let the liability fall on the defendant.
 I find those comments apposite.
It’s true that the passage, literally, states only that the trial judge thought the SCC had approved Haag. But the BCCA didn’t disagree – there or later.