Fig. 1 shows one participant’s specimens writing the sentence and producing superimposed circles at 18:00 hours of the first day and 03:00 hours. An evaluation of the legibility and uniformity of the handwriting specimens at these two times suggests that the sentence was legible at 03:00 hours just as well as at 18:00 hours (compare Fig. 1a and c). In general it has to be noted that specimens of handwriting of all participants were legible at any session. However, the kinematics of handwriting differed between sessions: the participant needed more time to write the sentence at 03:00 hours and the frequency of strokes was reduced at 03:00 hours (3.92 versus 4.58 Hz at 18:00 hours; Fig. 1a and c). The frequency of strokes when producing superimposed circles was also lower at 03:00 hours (3.68 versus 4.35 Hz at 18:00 hours; Fig. 1b and d). Comparing the pressure writing the sentence at 18:00 and 03:00 hours (Fig. 1a and c), the participant in the example of Fig. 1 exerted increased pressure at 03:00 hours; however, this result does not represent the group data. By contrast, pen pressure producing the superimposed circles was similar at 18:00 (2.86 N) and 03:00 hours (2.83 N) in line with the group data.
Figure 1. Specimens of writing the sentence at 18:00 hours (a) and 03:00 hours (c). Top: the trace of the handwriting (solid line: trace on paper; dotted line: trace in the air); middle: the corresponding vertical velocity (vy) as a function of time; bottom: pen pressure as a function of time. Producing of superimposed circles at 18:00 hours (b) and 03:00 hours (d). Left: the trace of circles; right: the corresponding vertical velocity as a function of time.
Download figure to PowerPoint
Fig. 2 shows the group data. The time course of the experimental data of handwriting performance (dotted lines) and the fitted sine curve (solid lines) are illustrated together with melatonin concentration and sleepiness measures (broken lines) obtained during the 40-h CR protocol.
Figure 2. Kinematics of handwriting and circadian phase markers as a function of time of day. Frequency of up- and down strokes per second writing the sentence (a) and the superimposed circles (b). Temporal regularity (standard deviation (SD) of stoke duration; c) writing the sentence (black square) and superimposed circles (white circles) as deviation from mean. The solid lines show the best fit between the experimental data (dotted lines) and the sine curve obtained from harmonic regression. Subjective sleepiness (d; KSS, 1 = extremely alert, 9 = extremely sleepy) and salivary melatonin concentration (e) combined in 3-h intervals. Error bars represent the standard error of the mean. Asterisks represent the second day of the CR. KSS, Karolinska Sleepiness Scale.
Download figure to PowerPoint
The duration of writing the sentence exhibited a circadian variation (F12,96 = 2.57, P < 0.01) with prolonged writing duration between 00:00 and 09:00 hours. The mean time needed to complete the sentence was 9288 ms (±223 ms). A statistically significant (P < 0.001) circadian rhythm was documented by harmonic regression with an amplitude of 224 ms (±2.4% of mean level) and the fitted peak phase at 17:37 hours.
The frequency of handwriting also revealed a clear circadian rhythm. The mean writing frequency showed a significant effect of session (F12,96 = 3.22, P < 0.01) but no interaction between session and task (F < 1). The circadian variation of the frequency in writing a sentence (Fig. 2a) and in producing superimposed circles (Fig. 2b) was simultaneous. At night the frequency of both tasks decreased, but re-increased until noon of the second day reaching approximately the same level as during the first day. The overall level of the frequency was similar for the two tasks [F < 1; 4.28 Hz (±0.12 Hz) writing the sentence and 4.32 Hz (±0.11 Hz) producing superimposed circles]. For the frequency of writing a sentence and circles, rhythms were statistically significant fitted by sine functions (P < 0.001). The fitted peak phases were noted at 15:35 hours (writing the sentence) and 15:27 hours (writing the circles). The amplitude of the circadian rhythm of writing frequency was 0.14 Hz for the sentence and 0.12 Hz for the superimposed circles (representing ±3.2% of the mean level for the sentence and ±2.7% for the circles).
There was no significant effect of session for stroke length (F < 1). While the tasks differed in script size (F1,8 = 12.12, P < 0.01) with smaller script size for the sentence (4.44 ± 0.11 mm compared with 7.00 ± 0.21 mm for the circles), there was no significant difference in the time course of the tasks (F < 1). The mean length of the trace on paper writing the sentence was 311.7 mm (±7.0 mm) and did not show a significant effect of session (F12,96 = 1.24, P = 0.270). Thus, script size did not significantly change during the CR.
The mean pen pressure that was exerted onto the tablet by the tip of the writing stylus did not show an effect of session (F12,96 = 1.11, P = 0.368) nor was the interaction between session and task (F < 1) significant. However, the effect of task revealed significantly higher pen pressure for the superimposed circles (2.24 ± 0.05 N) compared with the sentence (1.74 ± 0.07 N; F1,8 = 6.26, P < 0.05).
Movement fluency did not show an effect of session (F12,96 = 1.45, P = 0.171) nor an interaction effect of session and task (F < 1). The main effect of task (F1,8 = 26.17, P < 0.001) revealed higher movement fluency in the simple writing movements of producing superimposed circles (NIV: 1.04 ± 0.01) compared with the more complex movements of writing a sentence (NIV: 1.18 ± 0.02). The overall (task-specific) low values of NIV showed that the participants wrote and circled in a highly automated control mode.
Fig. 2c depicts the standard deviation of cycle duration for both tasks. The main effect of session of the two-factorial anova for both tasks was not significant (F12,96 = 1.29, P = 0.261). The main effect of task was significant (F1,8 = 533.24, P < 0.001) with a higher standard deviation of the period time for the sentence (68.5 ± 4.70 ms) than for the circles (13.1 ± 1.24 ms), which is due to the higher variability in writing a sentence. The interaction of session and task of the temporal variability was only a trend (F12,96 = 1.98, P = 0.085). However, the harmonic fit for the standard deviation of cycle duration writing the sentence was statistically significant (P < 0.01) and localized the fitted peak phase at 15:55 hours with an amplitude of 5.1 ms (representing ±7.4% of the mean level). For writing circles, no statistically significant fit could be found confirming the trend for an effect of task complexity on the circadian time course.
Circadian phase markers
Subjective sleepiness ratings (KSS) showed a circadian rhythm (F12,96 = 13.42, P < 0.001; Fig. 2d). The KSS values were quite high right from the start of the CR. The increased level on the second day reflects the effect of sleep deprivation during the CR. Salivary melatonin concentration also revealed a circadian rhythm (F12,84 = 20.37, P < 0.001; Fig. 2e), with an onset of averaged melatonin concentration (dim light melatonin onset, DLMO: >3 pg mL−1) in the late evening between 20:35 and 23:43 hours (22:17 hours ± 1:01 h).