Bipolar disorders are mental illnesses characterized by the recurrence of mood-episodes, that can have a severe impact on the life of individuals. The effect of lithium, one of the main medications used to treat acute episodes or prevent them from happening, is very different from one individual to an-other. So far, there is no way to predict in advance for whom patient this treatment will be effective or for whom it will not. Finding markers that can predict as early as possible the efficiency of this treatment is a major field of current research in psychiatry, in order to avoid maintaining an inefficient treatment for several years that can have negative side-effects. Over the past decades, it has been shown by multiple studies that lithium can act on the biological clock, that regulates circadian rhythmicity of the body (i.e. rhythms that presents a 24 hours periods, such as rhythms of sleep and activity, feeding, social activities...). But it is still very unclear whether the effect of lithium in regulating the mood in bipolar disorders is mediated by this action. Melatonin is one of the key-regulator of circadian rhythmicity of the human body. Our hypothesis, based on some previous studies, is that the action of lithium in type-1 bipolar disorder (BD-I) is related to an action on melatonin secretion. To test that, we want in this study to compare the noctunal secretion of melatonin between BD-I individuals with a good response to lithium versus with a poor response to lithium.