Early rising or delayed bedtime: which is better for a short night's sleep?

Eur J Appl Physiol (1987) 56:403--411 European Journal of

Applied Physiology and Occupational Physiology �9 Springer-Verlag 1987

Early rising or delayed bedtime: which is better for a short night's sleep?*

M. Clodor~ 1, O. Benoit r, J. Foret z, Y. Touitou 3, N. Touron I, G. Bouard 1, and A. Auzeby 3

U3 INSERM 47, Boulevard de l'H6pital, F-75651 Paris, Cedex 13 2 Laboratoire de Physiologie Neurosensorielle CNRS, 15 rue de l 'Ecole de M~decine, F-75270 Paris, Cedex 06 3 Laboratoire de Biochimie, 91, Boulevard de l'H6pital, F-75634 Paris, Cedex 14, France

Summary. The present study compares the effects on sleep and the subsequent period of wakefulness of delaying bedtime of 2 h or advancing rising time by 2 h in subjects clearly differentiated by morningness or eveningness in their circadian rhythms. Twelve young healthy good sleepers, six morning types (MT) and six evening types (ET), were selected. The data obtained from the second 24 h (night and day) with delayed bedtime (DB) and advanced rising time (AR) were compared with those obtained in the reference condition (R) with normal sleep schedules. Sleep was recorded polygraphically and rectal temperature was con- tinuously monitored during the nights and during the day following the second night of each condition. Subjective estimations of alertness, performance tasks and urinary steroids were analysed. Early rising appeared to be more disturbing than a late bedtime. The second shortened night showed fewer characteristics of recovery sleep in AR than in DB. The decrease in self rated alertness was a function both of the type of condition (DB or AR) and of the morning-evening typology of the subject. The largest decrease was observed in AR and in the ET subjects. AR also resulted in the most pronounced decrease in performance tasks and in an increase in urinary 17 ketosteroids without change in the 17 hydroxy-corticosteroids. The effects on rectal temperature were limited to short periods after bedtime in DB and rising time in AR.

Key words: Sleep -- Alertness -- Circadian -- Temperature -- Urinary steroids

Offprint requests to: O. Benoit at the address above

* This work was supported by grant DGRST no. 81/E/1390 et by CNRS RCP no. 080812

Introduction

Most studies of shift work comparing the reactions of workers to different shift times have con- cluded that the morning shift is accompanied by most complaints. In Europe, the morning shift generally starts early -- for instance, in France 22.4% of the workers start work before 0730 hours -- and this implies a very early rising. A simple explanation of the subjective opinions could be that they are essentially expressed by "evening people", that is by subjects who always have diffi- culty getting up early. But the frequency of subjective complaints about the morning being the worst shift cannot be accounted for only by inter individual differences.

Therefore in the present study, subjects clearly defined as morning or evening types were selected and they all submitted to 2-h curtailment of their sleep obtained by delaying bedtime in one series and advancing rising time in another. A 2-h time shift was chosen in order to limit the sleep deficit implied by such a protocol and because it mimics frequent real life situations requiring early or late working times. The aim of the study was to assess the differential changes induced by the two conditions in subjective alertness, in vigilance tasks and in physiological variables (body temperature and urinary corticoid hormones).

Material and methods

Subjects. Twelve healthy male volunteers (mean age 25 years, SD=2.3 years) were selected. They were good sleepers, sleeping normally for 7--8 h, and showed no obvious signs of im- paired diurnal alertness. They underwent a medico-psychological interview with special emphasis on their habitual sleep/ wake cycle, and had not taken any psychotropic drug during the preceding year. In addition, all subjects were given a bat-

404 M. Clodore et al.: Early rising or delayed bed time

tery of tests (MMPI, Cattell anxiety test and Eysenck Person- ality Inventory) and completed a 2-week sleep diary. Two groups of six subjects each were selected as Morning type (MT) and Evening type (ET) on the basis of the Horne and Ostberg questionnaire (1976) with five moderate subjects and one extreme subject in each group.

Protocol. The experimental protocol was designed to compare the two different conditions of limited sleep in relation to the habitual 8-h sleep routine as shown in Fig. 1. The three parts of the protocol, separated by 1 week with habitual sleep schedules, were: (1) the reference week (R): three consecutive nights with habitual sleep schedule; (2) the delayed bedtime period (DE): one reference night followed by two nights with a 2-h delayed bedtime; (3) the advanced rising time period (AR): one reference night followed by two nights with a 2-h advanced rising time. The last 24 h of each of the three conditions were spent entirely in the laboratory, In order to main- tain the sleep schedules in the laboratory as close as possible to the habitual ones for each group, bed and rising times were different for morning and evening types (Table 1). Naps were prohibited throughout the experimental period. Lights out and rising times were strictly applied. Experiments were conducted between October and March with the part including urine col- lection in six subjects (3 MT and 3 ET) being undertaken between January and March. The present study deals with the data obtained during the last 24 h of each condition, that is, during the second night of sleep reduction and the following day.

Sleep, self rated alertness, performance tasks, rectal temperature and urinary steroids were studied. Polygraphic sleep recordings were performed every night. Rectal temperature was measured with thermocouple probes and monitored con- tinuously when subjects were in the laboratory. Temperature data were stored every 5 rain.

Self-rated alertness was measured by means of the shortened form of the Activation/Deactivation Adjective Checklist of Thayer (1978), which comprises 20 adjectives. As proposed by the author, dimension A measured by factors General Acti- vation (GA) and Deactivation/Sleep (DS) is indicative of alertness, while dimension B given by the factors General Deacti- vation (GD) and High Activation (HA) is associated with subjective tension.

Table 1. Bed and rising times for the morning (MT) and evening type (ET) subjects in the three different conditions (R: reference, DE: delayed bed time, and AR: advanced rising time)

R DB AR

MT 2300 0700 0100 0700 2300 0500

ET 0000 0800 0200 0800 0000 0600

The Search and Memory (SAM) test of Folkard et al. (1976) was used for performance evaluation. This is a letter cancellation task with a maximal test duration of 6 min (2 rain for each of the three versions). The different versions consist of memorizing one, three and five letters and recognizing the first occurrence of one of the target letters. The test includes ten different blocks of each of the three versions. Prior to the experimental period, the subjects underwent training sessions on different days and completed all the ten different blocks of the three versions. Performance was described by two measures: the "accuracy score" which is the percentage of correct answers relative to the number of letters actually searched and the "performance speed" which is the average time taken to find a specific letter in each of the 30 lines.

Alertness ratings and SAM tests were performed evey 2 h from rising to 2000 hours at the end of the experimental day. Urine was collected at fixed times every 4 h throughout the 24 h. Levels of urinary 17 ketosteroids (17 KS) and 17 hydroxy-corticosteroids (17 0HCS) were measured using the reactions of Zimmerman (1935) and Silber and Porter (1954) respectively as well as dehydro-epiandrosterone (DHEA). Uri- nary creatinine was determined with the Jaffe reaction using a Technicon analyser, Steroid data were expressed in mg. g- of urinary creatinine.

Statistical analysis of data was carried out using variance analysis (ANOVA), Student's t-test and paired t-test. Two way analyses of variance completed by the Scheff6 contrast analysis were used to compare the three experimental conditions and the two groups of subjects. In addition, the time series of experimental data were analysed according to the single cosi-

1st w e e k

I 1 Reference

20 h 20 h R 20 h R 20

Delayed bed time 20 h R D B 1 D B 2 20

2 n d w e e k t ~ N ~ ' ~ ' I l i l l ' l l i n l l '

20 h R A R 1 Advanced rising time A R 2 2o

3 r d w e e k ' . ~ ~ ' ~ - ' ~ '

Sleep Sleep Sleep

Subjects confined in the lab L i

Fig. 1. Experimental protocol. The right part of each line represents the span of time during which subjects were confined in the laboratory. All the results of the present study (except sleep data) correspond to that period

M. Clodore et al.: Early rising or delayed bed time 405

nor method (Halberg et al. 1967) which allows estimation within 95% confidence limits of the parameters characterizing a circadian rhythm. These are: the period, here equal to 24 h since this figure corresponds to an average periodicity of the synchronization of the subjects by daytime activity and night- time rest; the acrophase or the peak time of the cosine function used to approximate the rhythm; the double amplitude equal to the total rhythmic variability in 24 h, i.e. peak to trough difference; the mesor (rhythm-adjusted mean) corre- sponding to the 24-h mean when sampling is performed at equal intervals as they were in the present experiment. The 24- h mean and the standard deviation are also presented. Com- puter programs and least squares method were used to find the best fitting sine function approximating all data. With this method a rhythm was validated when its amplitude differed from zero with p < 0.05.

Results

Sleep

Sleep in the reference nights was averaged across three nights and compared with the second night o f bo th sleep reduced condit ions. In bo th types o f night the sleep deficit and the length of the preceding wakefulness were similar. Moreover , there was no di f ference in sleep organizat ion be tween the two first nights of sleep reduct ion (DB and AR). Sleep in DB and AR condi t ions exhibi ted the characterist ics o f recovery nights: a shor ter sleep onset latency, less intervening wakefulness , some more slow wave sleep (SWS) and a little less R E M sleep than in the reference nights (Table 2). Despite the same degree of sleep reduct ion and the same length o f preceding wakefulness (18 h) there were some differences be tween the two condit ions: in DB, sleep was longer and both sleep la tency and wakefulness were significantly more reduced than in AR. The sleep eff iciency index was higher in DB than in AR, a di f ference which was more p r o n o u n c e d in evening than in morn ing type subjects (in ET, 0.95 vs 0.89 and in MT, 0.95 vs 0.94). As a rule evening subjects were more affected than morn ing types when they had to get up early.

Alertness

Table 3 gives the average level o f Thayer ' s subjective scales o f alertness, GA, DS, G D and H A in the three condi t ions. The alertness level decreased in bo th condi t ions o f sleep reduct ion. Significant differences were observed for factor G A in morning types and for factor DS in evening types. In bo th groups the d i f ference was due to R and A R condit ions.

Table 2. Mean values, with standard deviations, of the main sleep characteristics for reference nights (R) and for the second night of both conditions of reduced sleep (DB: delayed bed time and AR: advanced rising time). Statistical analyses among the three conditions: variance analysis ANOVA (df 2.27); comparison of DB et AR conditions: Student's t-test (df 11)

Average sleep characteristics under the three conditions

R DB AR A N O V A Stu- df 2/27 dent's

t-test

PST (rain) 440.2 344.4 331.3 F= 17.6 t=2.8 (4.3) (3.7) (3.5) p<0.001 p<0.02

SWS (min) 127.5 132 136.4 NS (6.5) (9) (8)

REM (min) 85 74 69.9 NS (2.8) (4.2) (3.9)

W (rain) 25.8 6.6 9.5 F= 8.5 NS (5.3) (1.3) (2.9) p<0.01

Stage 2 29.5 11.7 18.5 F=8.1 t=2.2 latency (3.7) (1.8) (3.1) p<0.01 p <0.05

First REM 80.6 76.3 65 NS latency (3) (6) (3)

Sleep 0.88 0.95 0.92 F =4.31 t=2.2 efficiency (0.002) (0.08) (0.02) p<0.023 p<0.05 index

The variat ions in H A level demons t ra ted that subjects felt more tense during the day after AR than after DB.

The diurnal evolut ion o f the G A / D S ratio is shown in Fig. 2. For bo th groups, there were no significant di f ferences in peak t ime among conditions. However , in the morn ing types differences among condi t ions were significant at 0800 and 1200 h. At 0800 hours, alertness level was lowest in the subjects going to bed late. At 1200 hours it was lowest in the A R condi t ion, which suggests an advance o f the "m id d ay dip". In contrast , in the evening types, the diurnal pat tern of G A / D S was the same in the three condi t ions with the two cond i t i ons o f r educed sleep resulting in an overall and similar decrease in bo th cases.

Performance task

The pe r fo rmance scores were averaged for each o f the six tests comple ted between 0800 and 2000 hours. Analysis o f the results o f SAM3 failed


Table 3, Mean values and standard deviations of the four factors in Thayer's Activation-Deactivation adjective checklist and mean score on the tension visual scale (calm-tense) calculated on the measures obtained from 0800 to 2000 as a function of morning/ evening type, in the three conditions (GA: general activation; DS: deactivation sleep; GD: general deactivation; HA: high activation. MT: morning type; ET: evening type; R: reference; DB: delayed bed time; AR: advanced rising time). Statistical analysis by one-way ANOVA (df2/179)

Thayer's factors

GA DS GD HA Tension analogue scale

R 14.8 8.8 14.3 5.8 4.5 (2.2) (1.8) (2.1) (1.3) (0.8)

DB 13.5 9.3 14.4 5.1 3.2 (2.7) (2.0) (2.2) (1.2) (0.9)

MT AR 13.4 9.0 14.4 5.4 4.7

(2.8) (2.5) (2.3) (1.5) (2.0)

ANOVA F= 4.1 NS NS F= 3.5 F= 5.8 p<0.05 p<0.05 p<0.02

R 13.9 9.0 13.7 5.7 3.7 (3.4) (2.2) (1.6) (1.8) (1.7)

DB 13.1 10.6 14.6 5.3 3.5 ET (4.3) (3.1) (1.5) (1.6) (1.5)

AR 13.1 10.8 14 5.9 4.2 (3.4) (2.5) (1.9) (1.9) (2.0)

ANOVA NS F= 8.6 NS F= 3.5 NS p < 0.001 p < 0.05

to show any significant t rend because o f the large inter- and intravariabili ty. Therefore only the results of SAM1 and SAM5, the easiest and the most difficult tests, will be presented.

SAM1. Because this test was easy to complete , the accuracy score was almost maximal at the beginning of the exper iment . In bo th groups o f subjects the pe r fo rmance speed was higher in the two sleep reduced condi t ions, AR being the bet ter (Fig. 3A). That d i f ference was significant for all the subjects (morning and evening types) ( F = 5.7, p < 0 . 0 1 ) (Fig. 3), a result which indicates that a training effect had increased pe r fo rmance speed.

SAM5. The results were similar for morn ing and evening types as shown in Fig. 3B. Therefore , the results for the 12 subjects were poo led for statistical analysis. As in SAM1, per fo rmance speed showed a t raining effect: it was highest in the 3rd week ( F = 12, p < 0.001). In contrast , the accuracy score was lowest in that same week (F--7 .1 , p < 0 . 0 1 ) . When t ime of day was considered, the

decrease was more p r o n o u n c e d at the end of the day (1600--1800 hours), a result which was significant bo th at 1800 and for the t ime span 1600-- 1800 hours for the evening group and for all 12 subjects taken together (F =4 .2 , p < 0.02 Fig. 4). Morn ing types showed the same t rend but the differences were not significant. In spite o f the increase in speed observed in the AR condi t ion, the pe r fo rmance level could not be main ta ined at that o f the R and DB condit ions.

Body temperature

No significant di f ference in rectal t empera tu re was found be tween morn ing and evening groups whatever the condit ion.

In bo th groups DB resulted firstly in a lower t empera tu re than in R at the beginning of sleep (equivalent to the t empera tu re af ter 2 h o f sleep) and secondly a t endency (p<0 .10) towards a lower t empera ture t rough than in the reference nights (Fig. 5).


I

co

o_ co

During the day, in AR the temperature increased rapidly after rising and remained above the curves of the R and DB conditions for about 3 h. Later, it did not differ significantly in the three situations (Fig. 5).

Urinary steroids

As shown in Table 4 the 24-h mean level of 17 KS varied as a function of the condition. AR resulted in a significant increase for the whole group (3 morning and 3 evening types). In contrast, the 24- h mean values of 17 OHCS were not affected by the conditions. DHEA mean values tended to be higher when sleep was reduced (particularly AR), but the difference did not reach the level of signif- icance. In all conditions, DHEA values reached 6%--7% of total 17 KS. Table 4 also gives the cosinor evaluation of the urinary steroids circadian rhythms. As a rule the acrophases were not modified by the conditions.

Discussion

Even though 2 h sleep deprivation on two consecutive nights had limited effects, these effects differed significantly between the two types of deprivation (DB and AR).

A = SAM 1

E T o'-1 loo , ~ MT

7"10 3 ~ ~2 ~ - ~ ' / <~ (D ~ ~ ~ 9590

610 3 i L I R DB AR

GA/DS M T

"E / ~ % = . " . "'~ A R

DB 7,

/ /

/ Z

i i i i

8 10 12 14 16 18 20 h

Time of day

GA/DS E T

>' R ~,~ ~ . . . . . . . . . DB

<

/ . $;,,,2 /;"

?, I ~ i i p i i p .

8 10 12 14 16 18 20 h Time of day

Fig. 2. A Morning-type subjects (MT). GA/DS values as a function of time of day in the three conditions: R (reference), DB (delayed bed time), AR (advanced rising time). At 0800 F = 2.7*: difference among conditions, significant at p < 0.05 At 1200 F=3.6"*: difference among conditions, significant at p <0.02. B Evening-type subjects (ET) same measurements

. . . . . . . . . ~ ET

- .= MT

R DB AR

k I I i

31o 3 i

219 3i

B = SAM 5 80

ET / M T /1 et

~ 75 e

) ~ 7o

r I I 65 R DB AR

\ \ \ \ \ \

\~\ \ \ \ \

\ \

MT

i i i

R DB AR

Fig. 3. Speed and accuracy score in the SAM1 and SAM5 tests in the three conditions (reference, delayed bed time, advanced rising time). '~ stands for the inverse of the average time necessary to actually read one letter till finding a target letter, expressed in m s - L "Ac- curacy score" stands for the percentage of target letters actually found

408 M. Clodore et al.: Early rfsing or delayed bed time

3�84184 Speed ( ms -1 )

SAM 5

/ , , A R /

/ .

,/ ~... /

/ \ . \ ,i �9 D B / "-o / /

/ / /

/ / / / /

i i i i i i

8 10 12 14 16 18 h

T i m e of d a y

9 0 A c c u r a c y sco re (%)

R

- \ / -,,,, DB

\ , \

' \

\

'~, A R

i i J i i

8 10 12 14 16 18 h

T i m e of d a y

Fig. 4. Time of day effect upon performance measured by SAM5. Averaged values obtained for the whole group of subjects. Standard deviation was approximately the same throughout the day respectively R_+0.021, DB+_0.022, AR+0.026 for speed and R_+6.18, DB_+4.6, A R + 5 . 2 for accuracy (see also Fig. 3)

37,5

p 37.0

c r

k13

D 3 6 . 5 uJ [32

36.0

AR

Lights o f f

( p - O , l O )

i

0

\

DB

i /f r i /

o < 0 0 5

I I I I

4 6 8 10

HOURS

:ii!! ill! ........... ~:~ii:i!~!!!!~!~ii~ii !

12 14 16 18

R

DB AR

Fig. 5. Rectal temperature curves obtained for the entire group of subjects in the three experimental conditions. Rec- tal temperature measurements have been averaged every hour from the usual time of lights off until the end of the experimental day. Standard error on the reference curve is given by the hatched area. When compared with the other two conditions, temperature tended to be lower at bedtime in DB than it was at usual bedtime (ddf 2/24 F = 2.7 p < 0.09). It was higher (ddf 2/22 F = 4.1 p < 0.03) after rising time in AR (7th and 8th hour)

Sleep data

A previous study has shown a similar organiza- tion of nocturnal sleep in morning and evening type subjects (Foret et al. 1985).

The sleep characteristics of both the second nights of sleep curtailment (DB or AR) showed changes related on the one hand to the short duration of sleep (less REM) and on the other to the influence of the previous sleep deficit (short sleep latency, decrease in intervening wakefulness, increase of both SWS amount and sleep efficiency). Some differences were found to be a function of sleep schedules: DB resulted in a shorter sleep onset latency, a smaller amount of wakefulness

and some more REM sleep than AR. Those results are in good agreement with Taub and Berger (1976), who found different sleep structures when sleep was advanced or delayed by 3 h. With the same duration, sleep was better when it started late than when it was interrupted early in the morning. In addition during the second night of AR, the sleep of evening subjects exhibited less pronounced hints of recovery than that of morning types. Evening people encountered more diffi- culties in recovering sleep when they had to get up early every morning as is frequently the case in real industrial conditions. Taub and Berger (1976) in a 3-h sleep reduction by delayed bedtime, as well as Horne and Wilkinson (1985) in an experi-


Table 4. Cosinor results. Comparative circadian rhythms of urinary steroids (17 KS, 17 OHCS and DHEA) in the three experimental conditions (R: reference, DB: delayed bed time and AR: advanced rising time) for the six subjects. The circadian rhythms were calculated from the urines collected every 4 h for 48 h in each condition

24 b Total variability Acrophase

mean _+ SD h �9 min

% R p

17 KS

R 6.85 a 2.92 a 43 b 13.30 40.6 NS +2.4

DB 8.12 3.36 41 11.06 51.9 NS _+ 1.2

AR 9.16 4.26 47 13.12 42.9 0.007 -+ 1.5

17 OHCS

R 4.28 3.04 71 15,31 52.1 0.016 + 0.74

DB 4.13 2.42 59 15.02 64.8 0.003 _+0.87

AR 4.19 3.64 87 15.19 75.4 0.001 +0.88

DHEA

R 0.46 0.42 91 16.29 45.2 0.025 _+0.29

DB 0.56 0.50 89 15.08 59.6 0.06 _+0.36

AR 0.64 0.42 66 15.05 37.9 0.05 + 0.24

Values given in mg �9 g-1 of creatinine Total variability was also expressed as percentage of the 24 h mean value. A significant difference was observed for 17 KS between the conditions ( F= 5.4 p < 0.01 df 2.10)

ment with some weeks of 2-h sleep reduction essentially by delayed bedtime, found about the same sleep characteristics as in the present DB condition, i.e. short sleep latency, increase in SWS, less stage 1 and wakefulness, less REM sleep. As schedules in the Home and Wilkinson study had been individually determined by the subjects, one may suppose that they were chosen as the best according to the individual's past ex- perience. The acceptability of a limited but chronic sleep reduction seems to be conditioned, at least to a large extent, by the free choice of sleep schedules. In contrast, the present results suggest that when sleep reduction is due to a very early rising time, as it frequently is in morning shifts, chances of a good tolerance are reduced in most individuals and are even worse for evening subjects.

Alertness

The four factors used (GA, DS, GD and HA) have been classified according to two dimensions: A, from activation to sleepiness; and B, from tension to placidity (Thayer 1978). In both dimen-

sions, the two factors of each pair, respectively GA-DS and GD-HA, are in general negatively correlated. Nevertheless, the present results show that experimental conditions do not affect the two factors of one pair to the same extent. For instance, for dimension B, only HA was modified by the experimental situation.

As expected, both types of sleep reduction resulted in a decrease in average value of dimension A, in agreement with Fort and Mills (1972); Her- bert et al. (1976) and Taub and Berger (1974). But comparisons with other studies is difficult because those studies have mainly put the emphasis on alertness at rising time and before bedtime, assessed in various ways (Taub and Berger 1976; Herscovitch and Broughton 1981; Home and Wil- kinson 1985 and Johnson and Mc Leod 1973). However, our results obtained at 0800 hours after all three sleep conditions are in good agreement with the post-sleep measurements in these other studies.

After 2 h sleep deprivation, the decrease in self-assessed alertness is a function both of the type of sleep reduction, DB or AR, and of the morning/evening typology of the subjects. The largest decrease in alertness was observed in AR


and in the evening subjects, and was associated with an increase in high activation, a (HA) factor regarded as sensitive to stress (Thayer 1978; Witt- maier 1974).

The SAM test is a task involving various loads on short-term memory: performance in the test has been shown to have a circadian evolution (Rutenfranz and Colquhoun 1979; Monk 1982; Monk and Leng 1982). The phase of its circadian rhythm is a function of memory load: in the case of a low load (SAM2) performance and body temperature are in phase while they are out of phase for the high load version (SAM6) (Folkard et al. 1976). The present results agree with these find- ings: SAM1 exhibited a circadian pattern with a peak time around 1400, while performance in SAM5 measured by the accuracy score was on a plateau between 0800 and 1800 hours after a normal sleep night, but tended to decrease through the day after sleep deprivation. AR resulted in the most pronounced decrease. In spite of the high memory load (SAM5), in all three conditions per- formances did not differ until 1400 hours but decreased in AR thereafter. In DB the performance was maintained at a somewhat lower level than in the reference condition, while in AR performance tended to decrease throughout the day: this decre- ment is probably related to the high load of the short-term memory component in the task, which is known to decrease in efficiency throughout the day (Folkard 1982). The lack of effect of sleep reduction on SAM1 performance can be accounted for by a strong practice effect and by the simplic- ity of the task.

Urinary steroids

Whether the sleep conditions could result in modification of the urinary elimination of steroids was examined. The data clearly show the absence of any modification in both the level and the general pattern of urinary 17 OHCS. In contrast, it was observed that 17 KS were significantly increased (by about 33%), especially in AR. The reason for such an increase could be related to the stress due to the experimental protocol. However, the lack of effect in the same subjects of this protocol on both 17 OHCS and DHA does not allow such an hypothesis. The elimination of steroids in the subjects in the AR condition in our study can be compared with that of shift workers starting at the morning shift. Reinberg et al. (1976) found significant increases in both 17 OHCS and 17 KS from the beginning to the end of the morning shift (4 days).

These data, though preliminary, could be due to effects on gonadal function of both sleep deprivation and early rising. But acute sleep deprivation of 24 h and 48 h have been shown to result in a decrease in circulating androgens in healthy subjects without a change of cortisol level in the second study (Akerstedt et al. 1980; Cortes-Galle- gos et al. 1983). Further work is needed to clarify this point with special reference to gonadal secre- tion.

All this confirms that early rising, especially for an "evening type", is more disturbing than going to bed late, as was suggested by the volume of complaint voiced about a morning shift.

Temperature

Acknowledgements. The authors wish to thank S. Folkard for constructive advice with the analysis of SAM data and S. Or- soni for revising the manuscript.

As previously shown (Foret et al. 1985) nocturnal evolution of body temperature is not related to the morning/evening typology of the subjects.

As to temperature, only the DB condition resulted in significant changes during the night. In agreement with Gillberg and Akerstedt (1982) temperature at sleep onset after a 2-h delay in bedtime did not differ significantly from what it would have been if the subject had been sleeping for 2 h. In other words, the effect of sleep per se cannot be determined from the circadian drop in body temperature. In contrast, when a subject went to bed later than usual, the effect of sleep (masking effect) could be observed and the temperature reached a level lower than that after the usual bedtime.

References

Akerstedt T, Palmblad J, De La Torre B, Marana R, Gillberg M (1980) Adreno cortical and gonadal steroids during sleep deprivation. Sleep 3:23--30

Cortes-Gallegos V, Castaneda G, Alonso R, Sojo I, Carraneo A, Cervantes C, Parra A (1983) Sleep deprivation reduces circulating androgene in healthy men. Arch Androl 10:33--37

Folkard S (1982) Diurnal variation in logical reasoning. Br J Psychol 66:1--8

Folkard S, Knauth P, Monk TH, Rutenfranz J (1976) The effect of memory load on the circadian variation in performance efficiency under a rapidly rotating shift system. Er- gonomics 19:479--488

Foret J, Touron N, Benoit O, Bouard G (1985) Sleep and body temperature in morning and evening people. Sleep 8:311--318


Fort A, Mills JN (1972) Influence of sleep, lack of sleep and circadian rhythm on short psychometric tests. In: Colqu- hour WP (ed) Aspects of human efficiency. English Uni- versities Press, pp 115--127

Gillberg M, Akerstedt TB (1982) Body temperature and sleep at different times of day. Sleep 5:378--388

Halberg F, Tong YL, Johnson EA (1967) Circadian system phase. An aspect of temporal morphology; procedures and illustrative examples. In: von Mayersbach H (ed) The cel- lular aspects of biorhythms. Springer, Berlin Heidelberg New York pp 20--48

Herbert M, Johns W, Murray E, Dore C (1976) Factor analysis of analogue scales measuring subjective feelings before and after sleep. Br J Med Psychol 49:373--379

Herscovitch J, Broughton R (1981) Performance deficits following short term partial sleep deprivation and subsequent recovery oversleeping. Can J Psychol 35:309--322

Horne JA, Ostberg O (1976) A self assessment questionnaire to determine morningness-eveningness in human circadian rhythms. Int J Chronobiol 4:97-- 110

Home JA, Wilkinson S (1985) Chronic sleep Reduction: daytime Vigilance Performance and EEG Measures of Sleepi- ness with particular reference to practice effects. Psycho- physiology 22:69--78

Johnson LC, Mac Leod WL (1973) Sleep and awake behavior during gradual sleep reduction. Percept Motor Skills 36:87--97

Monk Th (1982) The arousal model of time effects in human performance efficiency. Chronobiologia 9: 49-- 53

Monk Th, Leng VC (1982) Time of day effects in simple repe- titive tasks: some possible mechanisms. Acta Psychol 51:207--221

Reinberg A, Vieux N, Laporte A, Migraine C, Ghata J, Abulker C, Dupont J, Nicolai" A (1976) Ajustment de rhythmes circadiens physiologiques d'op~rateurs d'une raf- finerie, lors de changement d'horaires travail-repos tousles 3--4 jours. Arch Mal Prof 37:479--494

Rutenfranz J, Colquhoun P (1979) Circadian rhythms in human performance. Scand J Work Environ Health 5:167--177

Silber AH, Porter C (1954) Determination of 17,21-hydroxy- 20-ketosteroids in urine and plasma. J Biol Chem 210:923--932

Taub JM, Berger RJ (1974) Acute shifts in the sleep wakefulness cycle effects on performance and mood. Psychomatic Med 36:164--173

Taub JM, Berger RJ (1976) The effects of changing the phase and duration of sleep. J Exp Psychol 2:30--41

Thayer RE (1978) Toward a psychological Theory of Multidi- mensional Activation (Arousal). Motiv Emot 2:1--34

Wittmaier BC (1974) Test anxiety mood and performance. J Personal Social Psychol 29:664--669

Zimmerman W (1935) Eine Farbreaktion der Sexualhormone und ihre Anwendung zur quantitativen colorimetrischen Bestimmung. Z Physiol Chem 233:257--264

Accepted February 26, 1987

Documents

Early rising or delayed bedtime: which is better for a short night's sleep?