Henri Chateau occupies a distinctive position in the Roulette literature of the early twentieth century. He does not merely construct a progression, nor does he limit himself to commenting on probability theory: he attempts to organize frequencies, deviations, figures, virtual systems, capital and methods of exposure within a single framework. The result is a broad treatise, closer to a general theory of systematic play than to a simple betting recipe.
The core of the work is reconstructed through institutional bibliographic catalogs, quotations and reproductions available online. Since a complete and verifiable scan of the French edition is not available to us, we distinguish with particular care among confirmed bibliographic facts, theses directly attributed to Chateau and later reconstructions.
The work
La Science de la Roulette et du Trente-et-Quarante, published in Paris in 1926: 194 pages and 62 drawings by the author, according to the catalogs of the Bibliothèque nationale de France and the Médiathèque de Monaco.
The project
To describe the statistical laws of the game, examine known systems and organize a practice based on equilibrium, deviation, figures, fictitious players and differential play.
The modern assessment
A historically important and methodologically rich work. Its procedures alter selection, exposure and variance, but do not by themselves demonstrate a mathematical advantage over an ideal Roulette wheel.
The author and the context of the work
The Bibliothèque nationale de France identifies Henri Chateau as a French author who lived from 1870 to 1942. His interest in gambling was not incidental: three years after the treatise he published Maître du hasard, described in the Médiathèque de Monaco catalog as a “roman du jeu”. This helps place him in context: Chateau approached gambling both as an object of analysis and as a cultural and human phenomenon.
Much later German Roulette publications describe him as having served for fifteen years as a consultant to the Monte Carlo casino and as having examined more than one thousand systems. The claim is recurrent, but it is not confirmed by the institutional catalogs consulted; it should therefore be treated as an unverified bibliographic tradition, not as a certain biographical fact.
Why the book differs from a collection of systems
The Médiathèque de Monaco catalog classifies the volume under the subject “Roulette — Probability”. Later bibliographic descriptions emphasize the presence of numerous tables and graphic representations. Chateau therefore seeks to compare empirical observation with theoretical tendency before moving on to operational constructions.
La Science de la Roulette et du Trente-et-Quarante
The original edition was published in Paris by Éditions du Monde Moderne in 1926. The Monaco catalog records 194 pages in a 24 cm format; the BnF also notes 62 drawings by the author. Later German and Italian translations circulated in different forms and lengths, not always with reliable textual fidelity.
The title uses the word “science” in the sense typical of the gambling literature of the period: observation of permanences, calculation, classification of figures and rational construction of systems. It does not automatically amount to the experimental validation now required to demonstrate an advantage.
The three laws of numbers
Online reconstructions of the work organize its statistical section around three laws: equilibrium, deviation and figures. This is a useful distinction, provided it is not turned into the false idea that the game must immediately correct every imbalance.
1. Law of equilibrium
As the number of trials increases, relative frequencies tend toward theoretical probabilities. This is the domain of the law of large numbers. Chateau uses this convergence as a general framework, not as a guarantee that an overdue chance must appear on the next spin.
2. Law of deviation
In the short and medium term, absolute differences can widen even while relative frequencies converge. For Chateau, deviation is not an anomaly to be ignored, but a phenomenon to be measured and incorporated into the construction of play.
3. Law of figures
Sequences of Red and Black are broken down by length and configuration. Runs and alternations do not all have the same frequency; the question is whether their classification can provide information that is usable before the spin.
The permanence of 56,534 spins
Later Roulette sources attribute to Chateau the examination of one hundred Monte Carlo sessions, totaling 56,534 spins, comparing the observed number of runs of different lengths with the mathematical tendency. The reproduced values are very close to theoretical expectations. The figure is consistent with the book’s project, but it has not been possible to compare it directly with a complete scan of the French edition: it should therefore be regarded as a secondary reproduction, not a certified transcription.
From the permanence to fictitious players
Chateau’s most characteristic step is not to entrust the entire operation to one real player. The permanence is read through several fictitious players, each with its own record, rule and progression. These players do not add new physical events: they multiply the accounting viewpoints applied to the same sequence.
The so-called multiplication of events makes it possible to run several virtual systems in parallel without exposing the full sum of their stakes on the table. The operator observes opposing balances and bets only the difference. This is a sophisticated idea of compensation and exposure compression, but it does not produce independent events: all records derive from the same wheel spins.
Fictitious players
Separate virtual systems follow the same permanence with different rules or states. The plurality increases the possibilities for comparison, but also the risk of selecting after the event the player that performed best.
Reduction scales
Procedures intended to reduce or convert virtual exposures into a smaller real stake. Their operational value lies in stake control, not in creating new probabilities.
Net stake
When the virtual players require opposing stakes, only the difference is placed on the table. The operation avoids needlessly opposing bets, but the balance retains the expected value of the net positions.
Two compartments: equilibrium and deviation
In the practical formulation attributed to Chateau, the method is divided into two distinct compartments: one system works on manifestations of equilibrium and another on manifestations of deviation. Both simultaneously record Red and Black using slow progressions; the operator puts only the final differential into play.
The architecture is more important than the individual progression. Chateau seeks two mechanisms capable of behaving differently during opposite phases of the permanence, so that the weakness of one is compensated by the other. When a virtual player exceeds a certain loss, a rescue intervention is introduced on the opposite side, described in translations as a rescue Paroli.
The difference between compensation and advantage
Compensation can reduce the gross stake, distribute losses differently and make the capital profile more regular. It does not, however, eliminate zero or turn an independent event into a predictable one. To demonstrate an advantage, it would be necessary to show that the selection rule produces, on new data, a conditional probability higher than the one embedded in the payouts.
Capital, progression and the duration of a negative phase
Chateau assigns a central role to capital. Not because money changes the probabilities, but because a structure composed of several virtual players and slow progressions must withstand prolonged fluctuations. Insufficient capital ends the experiment before the mechanism can develop according to its rules.
Unlike D’Alost, who seeks an advantage through flat betting, rapporteurs and selection, Chateau accepts a very slow progression. The progression must be durable, not aggressive: the problem is not immediate recovery, but preventing a normal deviation from destroying the entire structure. Summaries of the work also report an operating limit of about five hundred spins for a negative phase, chosen partly for psychological reasons and for the sustainability of the session.
The real function of capital
To absorb variance, the cost of zero and temporary differences among the virtual records. More capital reduces the risk of ruin over the period observed, but does not turn negative expected value into positive expected value.
The function of the progression
To redistribute stakes over time and link the amount of exposure to the state of the system. Durability depends on the maximum level of the progression and on the capital available.
The function of the stop
To limit duration, decision fatigue and maximum loss. A stop changes the distribution of session outcomes, not the mathematical expectation per total unit wagered.
Chateau as a critic of systems
One of the most interesting aspects is that Chateau does not accept celebrated systems without reservation. A quotation preserved in the German translation observes that the book would be incomplete without discussing Jacques Inaudi and Charles Henry, but adds that it is impossible to guarantee that their procedures passed the test: Inaudi’s would have been tried on only about three thousand spins, a quantity judged insufficient, while Henry’s would have been applied only rarely.
This caution matters. Chateau belongs to the culture of “scientific Roulette”, but recognizes that a favorable result on a limited sample is not enough. His standard remains far from modern protocols — frozen rules, independent samples, statistical comparison and replication — yet it is more rigorous than much later promotional literature.
Chateau, D’Alost and Charles Henry: three different projects
Charles Henry
He seeks the sign of deviation in the short term through series of nine spins and a table of hypothetical laws. The focus is the temporary persistence of a configuration. Charles Henry analysis.
Theo D’Alost
He transforms the permanence through figures and fictitious rapporteurs, attempting to break the equilibrium of natural play without monetary progressions. Theo D’Alost analysis.
Henri Chateau
He combines observation of permanences, virtual players, compensation, differential betting and slow progressions. The focus is the overall organization of exposure.
The three authors share an attempt to find structure in the short and medium term, but they do not propose the same theory. Reducing all of them to a generic opposition between equilibrium and deviation would erase precisely the technical differences that make their works historically interesting.
Technical assessment under modern criteria
Contributions that remain useful
The distinction between relative frequency and absolute deviation; attention to figures; comparison between observed and theoretical values; capital control; netting of opposing positions; criticism of samples that are too small.
Elements not demonstrated
The ability of fictitious players to select favorable spins; the existence of a distinct predictive mechanism for equilibrium and deviation; the stable effectiveness of rescue progressions.
How to test it today
Define every virtual player, entry, exit, progression, stop and treatment of zero before the test; apply the rules to new permanences; measure result, risk of ruin and confidence intervals.
The decisive point
Multiplying records does not multiply information. If every decision depends on the same random permanence, an advantage can exist only if the rule fixed in advance identifies a real dependency or a non-random physical condition. Without that proof, the system remains a sophisticated historical construction of management and selection.
Henri Chateau’s legacy
Chateau’s value does not lie in having supplied a universally accepted solution to Roulette. It lies in having gathered within one architecture many problems that remain central: how large a sample is required, how a deviation is measured, what a figure represents, how opposing systems are compared, what capital sustains a procedure and where description ends and prediction begins.
For this reason, La Science de la Roulette et du Trente-et-Quarante remains a source of great interest to historians and serious students. It should be read without two opposite errors: neither as an infallible method guaranteed by the label “scientific”, nor as a worthless curiosity simply because its conclusions do not coincide with modern probability theory.
Sources and status of the reconstruction
Verified bibliographic data: Bibliothèque nationale de France and Médiathèque de Monaco. Reproductions of quotations from the German translation preserved on the Internet Archive and later Roulette reconstructions of the statistical and operational passages were also consulted. No claim is made to have read the complete French original: information that cannot be directly confirmed is expressly identified as attribution or secondary sourcing.