dealt _once_, we shall find it very difficult to show that there is any connection whatever between the colours of the cards corresponding to particular years and the number of spots on the sun’s face. But if we repeat the process a thousand times, we shall find certain instances among the number, in which red suits correspond to all the years when there are many spots on the sun, and black suits to all the years when there are few spots on the sun. If now we were to publish all such deals, without mentioning anything at all about the others which showed no such association, we should go far to convince a certain section of the public that the condition of the sun as to spots might hereafter be foretold by the cards; whence, if the public were already satisfied that the condition of the sun specially affects the weather of particular places, it would follow that the future weather of these places might also be foretold by the cards. I mention this matter at the outset, because many who are anxious to find some such cycle of seasons as Sir John Herschel thought might be discovered, have somewhat overlooked the fact that we must not hunt down such a cycle _per fas et nefas_. “Surely in meteorology as in astronomy,” Mr. Lockyer writes, for instance, “the thing to hunt down is a cycle, and if that is not to be found in the temperate zone, then go to the frigid zones or the torrid zone to look for it; and if found, then above all things and in whatever manner, lay hold of, study, and read it, and see what it means.” There can be no doubt that this is the way to find a cycle, or at least to find what looks like a cycle, but the worth of a cycle found in this way will be very questionable.[7] I would not have it understood, however, that I consider all the cycles now to be referred to as unreal, or even that the supposed connection between them and the solar cycle has no existence. I only note that there are thousands, if not tens of thousands, of relations among which cycles may be looked for, and that there are perhaps twenty or thirty cases in which some sort of cyclic association between certain meteorological relations and the period of the solar spots presents itself. According to the recognized laws of probability, some at least amongst these cases must be regarded as accidental. Some, however, may still remain which are not accidental. Among the earliest published instances may be mentioned Mr. Baxendell’s recognition of the fact that during a certain series of years, about thirty, I think, the amount of rainfall at Oxford was greater under west and south-west winds than under south and south-east winds when sun-spots were most numerous, whereas the reverse held in years when there were no spots or few. Examining the meteorological records of St. Petersburg, he found that a contrary state of things prevailed there.