There exists, finally, another kind of unconscious and involuntary fraud: that which is due to illusion. It is constantly found in spiritistic seances, where ninety-nine times out of a hundred mediums produce no real phenomena. They are, nevertheless, in earnest, but they do not take into consideration the rôle of memory and imagination. This is particularly the case with _intuitive_ writing mediums and ‘control’ mediums. With this order of phenomenon we rarely obtain verifiable indications; the ‘spirits’ utter plenty of commonplace generalities, but give no precise information. _Fraud_ is a misnomer in this case: being unconscious and involuntary, it cannot, correctly speaking, be called fraud; therefore it is better to reserve the word ‘illusion’ for it. I cannot think of analysing the question of fraud in detail. If examined closely it is extremely complicated. But, like Richet, I deem ‘it possible that in states bordering on trance, and in trance itself, the psychology of a medium may be very different from ours.’ I confine myself simply to indicating the result of my reflections, which are the fruit of a long series of observations. Let me renew my oft-repeated recommendation for avoiding fraud: Experiment with light, the greatest possible amount of light, and seek for simple phenomena, difficult, perhaps, to obtain, but easy to observe, such as raps and movements without contact. II. ERROR If I insist so much upon the necessity, especially in the beginning, of seeking only for phenomena observation of which is easy, it is because error of observation is facile. We need to be much accustomed to seances to be able to distinguish rapidly between probable phenomena and those which are certainly tricked. It is with this, as with everything else, a question of time and reflection. One of the causes of error, which it is highly important to avoid, is obscurity. For many simple phenomena darkness is unnecessary; therefore, from the very outset, we should exhort the personification to accept light. I have already frequently said that personifications are very suggestible. I know well it is not always so, and that at times the personification displays much obstinacy. Personifications of this class are especially observed with mediums who have long-acquired habits. It was so with Eusapia, who was only accustomed to giving dark seances. But even when the personification appears to have very decided ideas, it is possible, with a little ingenuity, to induce him to change. It is with them as with secondary personalities, or subjects to whom we have given a suggestion. We must enter right into the circle of suggested ideas in order to break it; it is a question of tact only.