A cooled CCD camera operated at −120°C with slow scanning mode read-out was used with a specially designed high-throughput lens system. The camera was placed in a light-tight room in complete darkness (schematic illustration of the experimental setup is shown in ). Five healthy male volunteers, in their 20′s, were subjected to normal light-dark conditions and allowed to sleep from 0:00–7:00. On the days of photon imaging, volunteers were kept in a room (400 lux) adjacent to the dark room. For imaging purposes, the body surface was wiped and the subject was left 15 minutes in the dark room for dark adaptation, after which the naked subject in sitting position was exposed for 20 minutes to the CCD camera. Measurements were carried out in every 3 hours from 10:00 to 22:00 and continued for 3 days. Just before and after the measurements, the surface body (thermography) and oral temperature were taken. Saliva was also collected after the photon measurements for the analysis of cortisol level as a biomarker of endogenous circadian rhythms. Temporal variation of photon emission intensity was calculated from image data with extraction of the face and body intensity.
Figure 1 A. Schematic illustration of experimental setup. B–F. Images of ultraweak photon emission from human body. B. Image of the subject under light illumination. C. Image at 10:10. D. Image at 13:10. E. Image at 16:10. F. Image at 19:10. G. Image at (more ...)
The daily variation of photon emission is shown in . In all images, photon emission intensity from the face was higher than from the body. Moreover, photon emission intensity from the face was not homogeneous: the central area around the mouth and the cheeks was higher than the lateral area and the orbits. Furthermore, the photon emission intensity on the face and upper body appeared to display time-dependent changes. We plotted total photon emission intensity over the body and face against time, averaged across the 5 volunteers (). Photon emission was weak in the morning, increased in the afternoon and peaked in the late afternoon (ca 16:00) (one way ANOVA, F4,74
4.10, P<0.005). These data strongly suggest that there is a diurnal rhythm of photon emission from the human body. To further support this conclusion, immediately following the end of the previous experiment three volunteers were kept awake in a light (400 lux) environment and photon emission was measured at 1:00, 4:00 and 7:00 AM (Supplementary Figure S1
). Photon emission formed a peak at late afternoon, then gradually decreased and stayed low at 1:00–7:00 AM in a constantly exposed light condition (400 lux), indicating the diurnal rhythm of photon might be caused by endogenous circadian mechanism.
Ultraweak biophoton emission was completely different from thermographic images showing surface temperature (). High photon emission were detected from the cheeks, followed by the upper neck and the forehead, while high temperature was detected in the supraclavicular lateral neck region, from which photon emission was low. In cheek, the highest level of emission reaches to 3000 photon/s·cm2 at 16:00 which is about double to the value at 10:00.
Next, we examined the correlation of photon emission to other physiological parameters known to show circadian variations. In the subject of , we found a temporal decrease of cortisol from morning to evening, in opposite to the increase of photon emission (). Cortisol concentration shows a clear daily rhythm, peaking in the morning and negatively correlated with photon emission intensity (p<0.002; from 5 volunteers; ). Body temperature, another parameter showing daily rhythms peaking at night, does not show significant correlation with photon emission (Supplementary Figure S2
Figure 2 A. Comparison of temporal variation of biophoton emission intensity and cortisol concentration in saliva observed through 3 days. Shaded regions indicate sleeping time. The subject is the one of B. Daily change of cortisol secretion (more ...)
The photon emission mechanism is thought to originate from the generation of free radicals in energy metabolic processes. The spectra of photon emission detected from the palm skin span from 500 to 700 nm, with primary and secondary emission peaks at 630–670 nm and 520–580 nm, respectively 
. Free radicals subsequently react with lipid or protein, generating electronically excited species as byproducts 
. These excited molecules, such as carbonyl group in excited triplet state from lipid peroxidation or proteins including excited tyrosine or tryptophan, can further react with fluorophores through energy transfer and lead to photon emission 
. Higher level photon emission on facial skin might be caused by differences in the content of melanin fluorophores 
between facial and thoracic skin.
No significant correlation of daily photon intensity and temperature was found, and the dissimilarity between photon emission and thermal image suggest that the diurnal rhythm of photon emission is not a consequence of a change of temperature or microcirculation. Moreover, a clear negative correlation of temporal changes of photon emission and cortisol might suggest that the diurnal rhythm of photon emission reflects the changes of cellular metabolic processes under the control of the circadian clock. Circadian rhythms are generated in most cells throughout the body, driven by clock genes interlocked in transcription/translation feedback loops 
. Recent advances of chronobiology have revealed that the redox state of the cells regulates circadian gene expression, indicating the importance of metabolic cues for clock oscillations 
. Indeed, glucose utilization, accompanied by oxygen consumption, shows robust rhythms in the main mammalian circadian center 
. By the regulation of cellular respiratory chain producing reactive oxygen species, which in turns react with molecules including proteins, lipids and fluorophores, whose excited states emit biophotons 
, the human body glitters to the rhythm of the circadian clock.