A certain number is observed not to have turned up for a considerable length of time, and the conclusion is formed that an increasing stake upon it must in the end prove a good investment. More often than not the very contrary is the case, for there have been whole days at Monte Carlo during which a number at one table has scarcely appeared at all. On the other hand, if a record of every coup at this table had been kept, the recurrence of every number would, in the course of time, be found to be practically the same. Complicated systems have often been devised, the main principle of which was covering a large proportion of the numbers, only a few, supposed by deduction to be unlikely to turn up, being left untouched. Disaster has invariably followed even a moderate run on such numbers, which, of course, occurs in the end, completely draining the players' pockets.

The even chances, without doubt, afford a player the greatest likelihood of success.

Staking a louis every time on both black and red, or any other even chance, leaving on any winnings in the hope of catching a run, is occasionally not a bad plan. The trouble of staking on both chances can be modified by calculation, though it is somewhat apt to lead to confusion.

A great number of players spend their whole time trying to strike a run at trente-et-quarante—this generally occurs when they are absent from their favourite table. The third coup would seem to be the most dangerous: for this reason, when a colour has run twice it is better to withdraw some portion of the sum staked, and then the remainder may be left to double up.

The practice of staking on the dozens at roulette is generally very attractive to those fresh to the tables, who like the idea of landing a two to one chance. The same type of player is, as a rule, at one time or another, fascinated by that system (or rather method of staking) which consists in backing two dozens, that is, laying two to one against the bank. Most of such players, however, soon discover how disastrous this may prove, and it should be realised that it is by no means an unusual occurrence for a dozen not to appear for ten or twelve coups—seventeen, I believe, is the record number of non-appearances. The great objection, however, to backing two dozens is zero, which sweeps everything but the even chances.

Another method of play is to stake against the recurrence of any number of even chances in an identical order.

Ten coups at trente-et-quarante, for instance, having resulted thus: