Sleep. 18(10):895-900
© 1995 American Sleep Disorders Association and Sleep Research Society
Environment and Sleep
Does a Subtropical Climate Imply a
Seasonal Rhythm in REM Sleep?
J.J .M. Askenasy and R. Goldstein
Sleep Medicine Institute, Sheba Medical Center,
Tel Hashomer, Israel
Summary: The present study was performed on 615 male subjects referred to the Sleep Medicine Center at Tel
Hashomer, Israel, and polysomnographically recorded for a single night between January I, 1990 and December
31, 1993. The study suggests the existence of a circannual rhythm of rapid eye movement (REM) sleep time with
an acrophase during December-January and a nadir during July-September (single cosinor analysis: mesor = 49.7
± 0.9, amplitude = 5.9 ± 1.2, p < 0.001). Both REM sleep time and REM sleep percentage were higher and REM
sleep latency shorter during winter and spring than during summer and fall. No dependence of the seasonal REM
sleep time rhythm upon age, apnea-hypopnea index or diagnosis type was detected. These data support, for a
subtropical climate, results previously obtained in a temperate climate. It is possible that external temperature may
be the principal factor influencing the phenomenon. Key Words: REM sleep-Seasonal variation-Circannual
rhythm-Male subjects-Subtropical climate.
Numerous experimental studies on the interrelationship between sleep architecture and climate have
shown that significant changes in rapid eye movement
(REM) sleep and slow wave sleep (SWS) parallel changes
in the external temperature (1-6) and that the nonrapid eye movement (NREM)/REM ratio depends on
the peripheral thermal state (7). Some studies suggest
that exposure to cold induces an increase in REM sleep
(8,9), whereas a rise in temperature increases SWS and
delays REM sleep (5,6,8). However, several studies
dealing directly with human sleep and external temperature contradict these findings; thus there is a diminished REM in polar conditions (3) and an increased
REM in a subsaharan climate (4).
Studies of seasonal sleep architecture rhythms (1,2,9),
which are correlated only indirectly to temperature,
are more common. Because the most conspicuous aspect of external temperature is its seasonal rhythm, all
the studies relating sleep to seasonal rhythms assume
that such rhythms are dependent on the external temperature. Weitzman et al. (10) did not find any significant season-dependent changes in sleep architecture.
A more recent study, however, described a significant
Accepted for publication June 1995.
Address correspondence and reprint requests to Dr. J. J. M. Askenasy, Sleep Medicine Institute, Building 16, Sheba Medical Center,
Tel Hashomer, 52621, Israel.
increase in REM sleep and a parallel decrease in SWS
during winter as compared with summer in a temperate
climate (2). In this longitudinal study, performed on
10 healthy young male subjects, Honma et al. described
seasonal variations in the timing of sleep and mean
body temperature (2). They concluded that the phaseangle difference of the rectal temperature rhythm and
sleep varied seasonally.
We hypothesized that if this conclusion were well
founded, a transverse analysis of sleep performed on
a large population over four consecutive years and during each of the seasons in our subtropical climate might
provide evidence of such a sleep variability. Our expectation was that if a sleep variable had a seasonal
rhythm, its interseasonal variability would exceed interindividual variability. A 4-year transverse study was
therefore carried out on a large number of patients,
each one undergoing a polysomnographic (PSG) examination for I night.
The seasonal variations in sleep architecture in male
subjects undergoing recordings in a sleep disorders center are presented here. It should be noted that the climate oflsrael is a mixture of two climatic types, "desert" and "rainy", consisting essentially of a long summer (hot and dry), a shorter winter (cold and wet),
short springs (warm and dry) and short autumns (mild
and dry).
895
J. J. M. ASKENASY AND R. GOLDSTEIN
896
TABLE 1. Main characteristics of the male population investigated over four consecutive years (1990-1993), by season
Population characteristics
Winter
Mean 47.1 ± 14.3
± SO
31.1%
Disease typc-% SAS
(103)
(number of
ED
38.1%
cases)
(91)
Other
40.0%
( 18)
Total number of cases
212
Age
Spring
Summer
Fan
Total
Test
p
46.2 ± 15.1
46.9 ± 14.1
47.7 ± 16.2
46.8 ± 14.9
ANOYA
0.89 (NS)
13.2%
(57)
20.5%
(49)
22.2%
36.3%
(120)
24.6%
(71)
20.0%
(9)
190
15.4%
(51 )
25.5%
(61)
17.8%
(8)
97
331
F(611,3)
(10)
116
=8
239
45
Multiple chi-square
test NS
0.10 (NS)
615
SO = standard deviation; NS = not significant; SAS = sleep apnea syndrome; ED = erectile dysfunction.
METHODS
Subjects
Only males were included in the present study. This
was because the large majority of patients referred were
men (all those with erectile problems, and most of
those with suspected sleep apnea), and the small number of women available would have introduced further
complications such as menstruation-dependent changes
in sleep pattern (11). Our study population therefore
consisted of 615 male subjects. The few who were on
medication when they were referred to us were told to
stop taking it for 8 days before PSG recording.
Each subject was polysomnographically recorded for
1 night between January 1, 1990 and December 31,
1993 in the sleep laboratories of the Sleep Medicine
Institute at Sheba Medical Center. The recordings were
carried out under stable indoor temperature conditions
of 18°-20°C. The mean age of the subjects was 46.8 ±
4.9 years (range 2-88 years). There were no statistically
significant differences in age in relation to season. The
subjects were referred as possibly suffering from sleeprelated breathing disorders (sleep apnea syndrome; SAS)
(n = 331), erectile dysfunction (ED) (n = 239) or other
problems (n = 45). There was no statistically significant
difference in disease distribution in relation to season
(Table 1).
Equipment and specific methods
A single-night PSG was performed by means of a
Nikon-Kohden Neurofax polygraph with 18 channels.
The full recording comprised three channels for electrooculography (EOG), two channels for electroencephalography (EEG), two channels for electromyography (EMG; submental and right tibialis anterior
muscles) and one channel for electrocardiography
(ECG), oxygen saturation, nasal air flow and chest and
abdominal respiratory movement transducers, respectively. The sleep stages were scored according to the
criteria of Rechtschaffen and Kales (12).
Sleep, Vol. 18, No. 10, 1995
The following sleep parameters were investigated: 1)
Expressed in minutes: time in bed, total sleep time,
sleep latency, REM sleep latency, stage I, stage II, stages
III-IV ofNREM sleep, REM sleep, longest awakening
and longest sleep period; 2) Expressed numerically:
REM periods, num ber of arousals, arousals longer than
5 minutes, movement periods and stage transitions; 3)
Expressed in percentages (out of total sleep time or
from total recording time): stage I, stage II and stages
III-IV of NREM sleep, total NREM sleep and REM
sleep; 4) Expressed as indices: sleep efficiency, apneahypopnea index (AHI) and O 2 desaturation.
Statistics
The data were statistically analyzed with EPI Info
Version 5 software. Because there were no statistically
significant differences in sleep with regard to year, the
data for all 4 years were summarized and analyzed
together in relation to season. The seasonal sleep variables represent the means of all the daily values obtained during the same season (winter, November 15March 15; spring, March IS-May 15; summer, May
IS-October 1; fall, October I-November 15) during
all 4 years together, so that the variance of these means
represents the seasonal interindividual variability.
The seasonal changes of numerical values were evaluated by analysis of variance (ANOV A) at a confidence
limit of at least I %. The seasonal changes in paradoxical sleep duration (REM sleep time = RST) according
to age group or to the AHI were estimated by multiple
analysis of variance (MANOVA) at a confidence limit
ofat least I %. Where significant differences were found,
the Newman-Keuls multiple comparison test (NKmct)
was used, at a significance level of at least I%. The
percentage values were analyzed by a multiple chisquare test.
Additionally, the monthly values (the cumulative
values of all days from the same month from all 4
years) of the sleep variables underwent a separate analysis for a circannual rhythm, using the mean cosinor
897
SEASONAL RHYTHM OF REM SLEEP
TABLE 2. Sleep measures, according to season, in male subjects investigated over four consecutive years (1990-1993)
ANOYA
Sleep measures
Winter
Sleep continuity measures
Total recording time (TRT)
(minutes)
Total sleep time (TST) (minutes)
Sleep latency (minutes)
Number of movements
N urn ber of arousals
Number of awakenings longer
than 5 minutes
Longest sleep (minutes)
Longest awakening (minutes)
Sleep efficiency (TST/TRT x
100)
NREM sleep measures
Stage I time (SI) (minutes)
Stage 2 time (S2) (minutes)
Stage delta (3 + 4) (SWS)
(minutes)
Stage I percent (SI/TST x
100)
Stage 2 percent (S2/TST x
100)
Stage delta percent (SWS/TST
x 100)
Total NREM percent (SI +
S2 + SWS)/TT x 100
REM sleep measures
REM latency (minutes)
REM time (RT) (minutes)
REM percent (RT/TT x 100)
Number of REM periods
352.2 ± 54.2
299.6
12.9
7.2
12.1
±
±
±
±
76.3
16.8
7.4
7.9
Spring
363.3 ± 65.7
306.4
10.6
8.7
13.4
±
±
±
±
68.5
11.5
6.9
9.9
F(614,3) =
P
366.5 ± 50.7
2.44
0.88 (NS)
±
±
±
±
71.8
13.4
6.3
6.4
3.02
1.41
1.12
1.36
0.06
0.24
0.34
0.20
Summer
Fall
364.5 ± 53.6
301.9
10.6
8.4
13.8
±
±
±
±
50.6
13.0
7.5
7.7
300.1
10.7
7.3
13.1
(NS)
(NS)
(NS)
(NS)
2.2 ± 2.0
110.3 ± 68.6
21.9 ± 19.7
2.0 ± 1.9
116.4 ± 64.8
24.3 ± 19.5
2.1 ± 1.7
112.1 ± 63.8
22.7 ± 20.0
1.2 ± 2.1
109.4 ± 64.2
25.7 ± 20.1
0.20
2.12
0.26
0.89 (NS)
0.10 (NS)
0.86 (NS)
81.7 ± 16.5
84.2 ± 12.6
85.3 ± 16.3
84.6 ± 16.6
0.75
0.53 (NS)
29.3 ± 26.9
168.3 ± 59.8
30.3 ± 25.4
172.1 ± 61.6
28.4 ± 22.3
172.4 ± 54.3
30.7 ± 72.7
171.2 ± 59.9
1.7
1.8
0.17 (NS)
0.16 (NS)
45.6 ± 21.7
50.0 ± 37.1
51.8 ± 41.7
48.9 ± 37.2
1.5
0.22 (NS)
9.8 ± 9.1
9.9 ± 8.2
9.4 ± 7.8
10.2 ± 9.2
2.0
0.12 (NS)
56.2 ± 14.8
55.2 ± 14.1
57.1 ± 13.6
57.0 ± 14.6
1.2
0.32 (NS)
15.2 ± 9.6
16.3 ± 12.6
17.2 ± 11.5
16.3 ± 11.5
1.6
0.18 (NS)
81.2 ± 13.6
82.4 ± 13.3
83.7 ± 14.4
85.6 ± 13.9
2.1
0.09 (NS)
109.6
54.4
15.3
3.2
±
±
±
±
52.2
25.3
6.5
1.3
111.8
52.6
14.1
3.1
±
±
±
±
63.0
23.7
6.3
1.3
124.6
46.3
12.9
3.1
±
±
±
±
71.7
29.6
5.7
1.3
99.8
41.5
13.5
3.0
±
±
±
±
62.7
30.2
6.7
1.3
4.12
4.14
5.1
0.9
<0.007
<0.007
<0.003
0.81 (NS)
Bold values represent the minimal values, statistically significant differing from non-bold values (Newman-Keuls multiple comparison
test at a significance level of at least 1%).
method. The analysis, which consisted of a single cosinor for all 4 years together, with a period of 365.25
days, made possible rhythm detection at a significance
level of at least 5%.
RESULTS
Of all the sleep variables analyzed, only the REM
sleep measures displayed significant seasonal differences (Table 2). Both RST and REM sleep percentages
were significantly lower during summer and fall than
during winter and spring (Table 2). To the contrary,
in comparison with spring and summer, REM sleep
latency significantly decreased during winter and fall
(Table 2).
No significant seasonal changes in sleep-related
breathing measures (respiration rate, AHI and O 2 desaturation) were detected (Table 3).
The fact that the MANOY A test, performed for a
seasonal RST pattern according to age groupings, revealed both significant seasonal and significant agerelated differences, whereas the age/season interaction
was not statistically significant, demonstrates that the
seasonal rhythm of RST is not age related (Table 4).
Not only did the AHI not display a significant seasonal pattern (Table 3), but analysis of the seasonal
TABLE 3. Sleep-related breathing measures, in male subjects investigated over four consecutive years (1990-1993), by
season
ANOYA
Sleep-related breathing
measures (± SD)
Winter
Spring
Summer
Fall
F(611,3)
p
Respiration rate
Apnea-hypopnea index (AHI)
O2 desaturation (%)
15.4 ± 3.4
20.2 ± 36.8
11.8 ± 7.8
16.2 ± 3.3
24.8 ± 36.6
10.7 ± 7.7
16.1 ± 3.4
23.8 ± 33.7
11.1 ± 8.3
16.0 ± 3.1
29.8 ± 35.1
11.3 ± 10.4
0.65
0.45
0.07
0.59 (NS)
0.72 (NS)
0.97 (NS)
Sleep, Vol. 18, No. 10, 1995
l
J. J. M. ASKENASY AND R. GOLDSTEIN
898
TABLE 4.
Mean REM sleep time (RST) in male subjects investigated over four consecutive years (1990-1993), by season
and age group (± SD)
Age group (years)
21-40
<20
Season
41-60
>60
42.5 ± 21.2 (40)
51.3 ± 24.7 (119)
56.0 ± 26.4 (44)
67.8 ± 37.3 (9)
Winter
41.5 ± 21.8 (21)
51.2 ± 22.3 (62)
60.6 ± 31.2 (8)
56.9 ± 23.7 (25)
Spring
45.2 ± 20.5 (99)
42.3 ± 19.4 (50)
42.0 ± 17.0 (35)
55.5 ± 31.1 (6)
Summer
40.5 ± 21.7 (17)
47.0 ± 23.4 (54)
35.0 ± 26.4 (22)
50.3 ± 31.7 (4)
Fall
Numbers in parentheses represent numbers of subjects. MANOVA: F(599,3) = 4.39, p < 0.001 for season factor; NKmc test p = 0.01,
Winter = Spring> Summer = Fall, F(599,3) = 3.51, p < 0.001 for age factor, NKmc test p = 0.01 «20) > (21-40) = (41-60) > (>61),
F(599,9) = 0.23, NS for season-age interaction.
changes in RST according to AHI values produced
evidence that these changes were not dependent upon
the AHI index (Table 5).
REM sleep time (RST) values were significantly
higher in subjects with SAS and ED than in subjects
with other diagnoses, but because no significant diagnosis/season interaction was detected it can be stated
that diagnosis did not affect the seasonal pattern of
RST (Table 6).
When we analyzed the data not on a seasonal but
on a monthly basis, cosinor analysis confirmed a circannual rhythm only for RST. Between January 1, 1990
and December 31, 1993, the monthly mean variations
of RST displayed a circannual rhythm whose acrophase was during December and January and nadir
during July, August and September (Fig. 1). The mean
cosinor analysis showed that this rhythm was statistically significant (mesor = 49.7 ± 7.0; amplitude =
5.9 ± 1.2; phase = -6 ± 11; p < 0.001). Of interest
is the fact that, as an exception to this general pattern,
there was a sharp RST decrease during March (Fig. 1),
when a transitory increase of the external temperature
is felt in the Israeli climate as a result of the hamsin
(a very hot and dry desert wind).
DISCUSSION
The present results show that, at least in male subjects referred to a sleep disorders center and under the
subtropical climatic conditions of Israel, REM sleep
duration or time (RST) displays significant seasonal
differences, with high values during the winter and
spring and low values during summer and fall. The
RST also displays a circannual rhythm that reaches its
maximum in December and January and its minimum
from July to September. Because our study was carried
out for one single night on a large population of different subjects, the present results can be interpreted
as showing that there is a general tendency for the
interseasonal differences of RST to exceed the interindividual differences.
The fact that in a subtropical climate both RST and
REM sleep percentages are high during winter and low
during summer, whereas REM sleep latency displays
an inverse pattern, provides evidence for a winter increase (or a summer decrease) in REM sleep pressure,
completely confirming results previously obtained in
a temperate climate (1,2).
Factors known to affect REM sleep, such as age (13),
AHI (14) or type of diagnosis (15), did not significantly
influence the seasonal pattern of REM sleep that we
have noted, and sleep-related breathing disorders did
not display any significant seasonal rhythm. These
findings strengthen the supposition that REM sleep
displays a season-dependent rhythm and support the
idea that, apart from all factors that could possibly
interfere with REM sleep, this rhythm is a general characteristic of male subjects. (Our study, as stated, excluded women.) It is evident that none of these factors
could of itself induce a different sleep structure at different times of the year, because the various presenting
problems and ranges in age and AHI were equally dis-
TABLE 5. Mean REM sleep time (RST) (± SD) in male subjects investigated over four consecutive years (1990-1993), by
season and age group
Apnea-hypopnea (AHI) group
Season
Winter
Spring
Summer
Fall
<10
51.0
55.5
46.0
39.0
±
±
±
±
23.6
25.3
20.5
21.7
10-20
(105)
(63)
(91)
(54)
20-40
>40
55.4 ± 25.3 (31)
56.0 ± 25.8 (40)
54.3 ± 29.2 (36)
51.6 ± 23.5 (19)
57.6 ± 21.7 (21)
45.6 ± 20.7 (13)
50.1 ± 22.8 (32)
46.3 ± 16.5 (31)
42.6 ± 16.3 (36)
45.5 ± 22.8 (8)
41.0 ± 22.0 (19)
40.6 ± 19.7 (16)
Numbers in parentheses represent numbers of subjects. MANOV A: F(599,3) = 6.37, p < 0.001, for season factor; NKme test p = 0.01,
Winter = Spring> Summer = Fall, F(599,3) = 0.22, NS for AHI group factor, F(599,9) = 0.43, NS for season-NHI interaction.
Sleep, Vol. 18, No. 10, 1995
899
SEASONAL RHYTHM OF REM SLEEP
CIRCANNUAL RHYTHM OF RST
(Data lire cum ulatod for 1990·1993)
Q
(/)
-!..
+
:E
=:
~ 52
(/)
It
48
2
FIG. 1.
3
Circannual rhythm of RST. RST
4
=
5
6
7
MONTHS
8
10
11
12
REM sleep time (duration).
tributed throughout the seasons, and the peculiar relationship between season and REM sleep cannot possibly be explained in terms of individual age or pathology.
The present data about RST are apparently in divergence with those from previous studies performed
in extreme climatic conditions, in which it was found
that RST diminishes in a polar climate (4) and increases in subsaharan conditions (5). This discrepancy
may be explained in one of two ways. Studies (4) and
(5) may provide evidence of the stressful effect of extreme temperature itself rather than of a true seasondependent rhythm; or we may suggest that because the
weather in Israel is mainly warm for 10 months ofthe
year, people are not accustomed to the cold conditions
of winter and therefore display a marked response to
them during the months of December and January.
Because the temperature within the sleep laboratory
was constant throughout the year, the response is clearly not sparked off by a chance coolness for a single
night.
Our transverse study confirms the increase of REM
duration in winter that has been observed in a longitudinal study performed in a temperate climate (2), as
well as the increase in REM sleep observed after exposure to cold air (8,9). The entrainer mechanisms of
the seasonal rhythm of REM sleep are unknown. However, from the factors known to be possible REM sleep
synchronizers, only two can be relevant here: external
temperature (2) and day length in relation to light intensity (16). The fact that the occurrence of REM sleep
is related to a decrease in core body temperature (7)
attests to a possible involvement of temperature in
synchronizing REM sleep, as has been suggested in
TABLE 6. Mean REM sleep time (RST) (± SD) in male subjects investigated over four consecutive years (1990-1993), by
season and diagnosis type
Season
Sleep apnea syndrome (SAS)
Erectile dysfunctions (ED)
Other diseases (OTH)
53.0 ± 26.8 (I8)
48.5 ± 21.8 (91)
61.8 ± 26.6 (!O3)
Winter
47.2 ± 23.8 (49)
54.2 ± 27.3 (10)
56.4 ± 23.2 (57)
Spring
38.9 ± 17.9 (61)
51.6 ± 22.4 (9)
48.5 ± 21.2 (120)
Summer
33.9 ± 21.5 (38)
42.5 ± 29.7 (8)
48.0 ± 26.5 (51)
Fall
Numbers in parentheses represent numbers of SUbjects. MANOYA: F(603,3) = 4.96, p < 0.001 for season factor; NKmc test p = 0.01,
Winter = Spring> Summer = Fall, F(603,2) = 6.89, p < 0.001 for diagnosis type factor, NKmc test p = 0.0 I, SAS = OTH > ED, F(603,6)
= 0.61, NS for season-diagnosis interaction.
Sleep, Vol. 18, No. 10. 1995
J. J. M. ASKENASY AND R. GOLDSTEIN
900
connection with the regulation of the NREM/REM
ratio (lO). The sharp decrease in RST observed during
March in Israel, when there is a transitory increase in
external temperature but the days change very little in
length, strengthens the suggestion that an increase in
external temperature is immediately followed by a decrease in REM sleep. However, we are also inclined to
believe that day length in relation to light intensity
may well play an important role in RST seasonal variations.
In conclusion we may suggest that in the subtropical
climate ofIsrael and in male subjects referred to a sleep
center, REM sleep increases during the winter and decreases during the summer months. Additionally, we
suggest that when REM sleep duration and REM percentage are analyzed for any reason, the "normal ranges" should take into account not only age, sex, etc.,
but also month or season.
Acknowledgements: We thank Professor Erhard Haus for
providing the mesor-cosinor analysis of our data, and we
!hank the Israel Meteorological Service for supplying climatic
data.
REFERENCES
I. Kohsaka M, Fukude N, Honma K, Honma S, Monta N. Sea-
sonal variatiort in the human circadian rhythm. Sleep Res 1991;
20A:542.
2. Honma K, Honma S, Kolisava M. Seasonal variation in the
human circadian rhythm association between sleep and temperature rhythm. Am J Physiol 1992;262:885-91.
3. Buguet A, Rivolier J, Jouvet M. Human sleep patterns in Antarctica. Sleep 1987;10:374-82.
Sleep, Vol. 18, No. 10, 1995
4. Buguet A, Hankourao 0, Gati R. Self-estimates of sleep in African students in a dry tropical climate. J Environ Psvchol 1990'
.
,
10:363-9.
5. Home JA, Staff LFE. Exercise and sleep: body heating effects.
Sleep 1983;6:36-46.
6. Bunnell DE, Agnew JA, Horvath SM, Jopson L, Willis M. Passive body heating and sleep: influence of proximity to sleep.
Sleep 1988; 11:210-9.
7. Heller HC, Glotzbach SF. Thermoregulation and sleep. In: Shitzer
A, Everhardt RC, eds. Heat transfer in medicine and biology:
analysis and applications. Vol. 1. New York: Plenum, 1985;
107-34.
8. Muzet A, Erhart J, Candas V, Libert JP, Vogt JJ. REM sleep
and ambient temperature in man. Int J Neurosci 1983; 18: 11726.
9. Galland BC, Peables CM, Bolton DPG, Taylor BJ. Sleep state
organization in the developing piglet during exposure to different
thermal stimuli. Sleep 1993;16:610-9.
10. Weitzman ED, Degraaf AS, Sassin JA, Hansen T, Godtlibsen
GR. Acta Endocrinol (Copen h) 1975;78:63-76.
1 I. Bamford CR. Menstrual-associated sleep disorder: an unusual
hypersomniac variant associated with both menstruation and
amenorrhea with a possible link to prolactin and metoc1opramide. Sleep 1993; 16:484-6.
12. Rechtschaffen A, Kales A. A manual of standardized terminology, techniques and scoring system for sleep stages in human
subjects. NIH Publication no. 204. Washington DC: U.S. Government Printing Office, 1968.
13. Carskadon MA, Orav JE, Dement We. Evolution of sleep and
daytime sleepiness in adolescents. In: Guilleminault C, Lugaresi
E, eds. Sleep/wake disorders: natural history, epidemiology, and
long-term evolution. New York: Raven Press, 1983:201-16.
14. Witting RM, Romaker A, Zorick FJ, Roehr TA, Conway WA,
Roth T. Night-to-night consistency of apneas during sleep. Am
Rev Respir Dis 1984; 129:244-6.
15. Carskadon MA, Rechtschaffen A. Monitoring and staging human sleep. In: Kryger MH, Roth T, Dement WC, eds. Principles
and practice ofsleep medicine. Philadelphia: WB Saunders, 1989:
665-83.
16. Borbely AA. Effects of light on sleep and activity rhythms. Prog
NeurobioI1978;10:1-31.