However, it appears very improbable that the bank would run such a risk. It is true that detection of roguery is not easy, where the tables are in a principality under an absolute monarch, and where police and every authority are interested in the continuance of the gambling. There is, however, the risk of some croupier “giving away the show”; and there is also the risk of detection. But—is cheating necessary? Is it worth its salt? Let us look closer into the acknowledged system. While playing on the even chances gives 1·35 per cent. in favour of the bank, playing on any other gives the bank 2·70; and as many fools play on those chances that favour the bank most highly, it is probably safe to assume that the odds in favour of the bank will average 1·66 on all the tables, both trente-et-quarante and roulette. If individuals playing would take in _all_ the money they could afford to lose, divide this into so many maximums (if one did not suffice) and stake the full maximum on each chance, and then retire, whether winners or losers, they would then have given the bank the least possible advantage, as they would have subjected themselves to the chances of the zero appearing the least possible number of times. As, however, almost every player wishes to have as long a run for his money as possible, almost all players, whether playing by a so-called system or not, divide their stakes, whether made on an increasing or on a decreasing scale, or haphazard, into a number of comparatively small stakes, so as to stay in the game as long as possible, with the result that the bank’s percentage is constantly working against them. The thinner they spread out their money, and the longer they stay in the game, the greater are the chances of their losing their money.