=65. Hygrometers.= While there are instruments designed to indicate the humidity by means of a hygroscopic substance, not one of them seems to be of sufficient accuracy for use in ecological study. The difficulty is that the hygroscopic reaction is inconstant, rather than that the instruments are not sufficiently sensitive. A number of hygrometers have been tested, and in all the error has been found to be great, varying usually from 10–20 per cent. In the middle of the scale they sometimes read more accurately, but toward either extreme they are very inexact. It seems probable that an accurate hygrometer can be constructed only after the model of the Draper psychrograph. Its weight and bulk would make it an impossible instrument for field trips, and the expense of one would provide a dozen psychrometers. In consequence, it does not seem too sweeping to say that no hygrometer can furnish trustworthy results. Of simple instruments for humidity, the psychrometer alone can be trusted to give reliable readings. Crova’s hygrometer, used by Hesselmann, is not a hygrometer in the sense indicated. As it is much less convenient to handle and to operate than the cog psychrometer, it is not necessary to describe it. _Psychrographs_ [Illustration: Fig. 7. Draper psychrograph.] =66. The Draper psychrograph.= A year’s trial of the Draper psychrograph in field and planthouse has left little question of its accuracy and its great usefulness. Essentially, it consists of a band of fine catgut strings, which are sensitive to changes in the moisture-content of the air. The variations in the length of the band are communicated to a long pointer carrying an inking pen. The latter traces the record in per cent of relative humidity on a graduated paper disk, which is practically the face of an eight-day clock. The whole is enclosed in a metal case with a glass front. A glance at the illustration will show the general structure of the instrument. Continued psychrometric tests demonstrate that the margin of error is well within the efficient difference for humidity, which is taken to be 5 per cent. In the field tests of the past summer, two psychrographs placed side by side in the same habitat did not vary 1 per cent from each other. The same instruments when in different habitats did not deviate more than 1 per cent from the psychrometric values, except when the air approached saturation. For humidities above 90 per cent, the deviation is considerable, but as these are temporary and incident upon rainfall, the error is not serious. For humidities varying from 10–85 per cent, the psychrograph is practically as accurate as the psychrometer. Per cents below 10 are rare, and no tests have been made for them. [Illustration: Fig. 8. Instrument shelter, showing thermograph and psychrograph in position.]