As the muscle-record and the time-record are separate, certain error is likely to be introduced in inferring the time-value of any point on the muscle-curve (fig. 64). This error becomes relatively serious when the total time to be measured is very small. There is, again, the difficulty of exactly determining the point of flexure which represents the beginning of mechanical response. More troublesome still is the error due to the inertia of the recording-lever. On account of this and the mechanical inertia of the responding muscle itself, the latent period thus obtained appears somewhat in excess of the true value.
![](http://upload.wikimedia.org/wikipedia/commons/thumb/0/02/Researches_on_Irritability_of_Plants%2C_fig._64.png/220px-Researches_on_Irritability_of_Plants%2C_fig._64.png)
In the apparatus which I employed, these difficulties have been reduced to a minimum. In the first place, the curve of response or phytogram is at the same time a chronogram. The error which might arise from an inference based on a neighbouring time-record is thus eliminated. I will later explain also the means that make it possible to determine the point of flexure, representing the beginning of the responsive movement, with relative accuracy. And lastly, the error due to the inertia of the recording part of the apparatus is reduced to a minimum by making the writing-lever excessively light. In the muscle recorders the weight of the recording-lever is about 3.5 grams. The lever which I employ weighs only .04 gram. The recording part of my apparatus is thus nearly a hundred times lighter than that used for muscle records.
The accuracy of the time-record when made by the response recorder itself may be gauged from records giving simultaneous tracings of the exciting standard tuning-fork