The older writers mention a typus inversus by which they understood a reversal of these symptoms-that is, first, sweating; then fever; and finally the chill. I have never seen such an occurrence, and in more recent communications I have found nothing further in relation to it. It seems to me, therefore, that this "type" was theoretically constructed, and not the result of observation. The so called "febris dissecta"-that is, with an interval between the three stages-I have likewise never observed, and doubt its occurrence. The duration of the whole paroxysm is variable, and may fluctuate between four and forty eight hours. Corresponding to this, the paroxysms are divided into short, moderately long, and long. Laveran, for instance, reckons the paroxysms of four to eight hours' duration among the short; of eight to twelve, among the moderately long; and over this, among the long. The duration of the paroxysm is closely connected with the evolution of the parasite, and will be considered in its proper section.

Lorain, who deserves credit for his thermometry of malarial fever, determines the relation of the three stages to one another in the figures 3 : 2 : 7. The longer the paroxysm lasts, the longer is the hot stage, though its period is always the shortest. Griesinger reports the rare occurrence of paroxysms extending over three days, in which the chill occupied the first day, the other two stages the second and the third.

The configuration of the fever curve, as might be judged, is not always the same. In short paroxysms the curve runs to a point; in long ones we see more or less extended levels. Moreover, the curve does not always show a straight rise and fall, though this may be characterized typical; the decline is not frequently interrupted by a new rise, which may even be followed by a second and a third before the final critical fall occurs.

To determine the beginning of the paroxysm the thermometer is necessary, because the chill, in a large majority of cases, breaks out one to several hours afterward. The assertions of the patients, therefore, as little as they are to be depended upon in relation to the time of the chill, are even less trustworthy for the accurate determination of the beginning of the paroxysm. This can be learned only by the rise in temperature.

According to Maillot and the majority of observers since him, most paroxysms (about two thirds of all cases) occur between midnight and midday-in other words, in the morning. Moreover, according to Maillot, the greatest number of quotidian and tertian paroxysms occur about 10 a. m. ; the smallest number, between 9 p. m. and midnight.

In my experience of 107 cases 91 per cent, of the paroxysms (from the standpoint of the commencement of the temperature rise) occurred during the period between 10 a. m. and 3 p.m.; in 6 cases the paroxysms occurred early in the morning; in 5 in the evening, and only in 3 between 10 p. m. and midnight. My experience, therefore, corresponds to Maillot's. Maurel observed in Guiana the majority of paroxysms between 2 and 5 p. m.