Physiology & Behavior 179 (2017) 208212

Contents lists available at ScienceDirect

Physiology & Behavior

journal homepage: www.elsevier.com/locate/physbeh

Calibration and validation processes for relative humidity measurement by a MARK

Hygrochron iButton
Mirim Shina,, Raymond Pattona, Trevor Maharb, Angus Irelandb, Paul Swanb, Chin Moi Chowa
a
Exercise, Heath & Performance Research Group, Faculty of Health Sciences, The University of Sydney, Australia
b
Australian Wool Innovation Limited, The Woolmark Company, Australia

A R T I C L E I N F O A B S T R A C T

Keywords: Accurate relative humidity (RH) measurement is demanded in studies of thermal comfort. Thermal discomfort
iButton occurs when the near-to-skin temperature or RH is outside of the thermal comfort zone. The Hygrochron, a small
Calibration wireless device which measures both temperature and RH, would be suitable and convenient in exercise or sleep
Relative humidity measurement studies. However, the RH measurement has not been validated. This paper has three parts. Part 1: In evaluating
Validation
the sensor surface for RH detection, four Hygrochrons were placed on a wet paper towel. Two were placed on the
Reproducibility
towel with the protruding surface facing up and the other two facing down. The results showed that the
Hygrochron with the protruding side was the sensor surface for detecting RH. Part 2: Twenty-seven Hygrochrons
were calibrated in a humidity calibration chamber at a RH range from 40 to 90% at a constant temperature from
32 to 37 C. The mean bias was 1.08% between the Hygrochrons and the calibration chamber. The
Hygrochron overestimated RH at the lower range (4060%) and underestimated RH at the higher range
(8090%). The application of individual regression equations to each Hygrochron improved accuracy and re-
duced the mean bias to 0.002%. However, one Hygrochron showed outlier values that may be due to a
manufacturing defect. Part 3: The reproducibility of Hygrochron for RH measurements was tested twice at the
same condition of 35 C over a three-month interval. The intra-class coecient was 0.996 to 1.000 with non-
signicant dierences in the mean RH between test and re-test results (p = 0.159). Hygrochrons are valid for RH
measurements which show high reproducibility. It is recommended that Hygrochrons be calibrated over a range
of desired RH and temperature prior to use to improve accuracy and detect any manufacturing defects.

1. Introduction than the manufacturer's specication for accuracy of 0.5 C. Fol-

The iButton (Maxim Integrated, San Jose, CA, USA) is a small
(16.25 5.89 mm2), rugged and self-sucient temperature and/or
relative humidity (RH) measuring device, designed to monitor tem-
perature-sensitive consignments such as food and chemicals during
transport. On account of its accuracy, low-cost, comfort and toler-
ability, it has been employed in scientic research including exercise
testing [13]; circadian rhythm and sleep research [48]; monitoring
participants' compliance and comfort in medical appliance use [911]
and in clinical diagnostic research [12].
The majority of the published literature reported the use of the
Thermochron (DS1921/DS1922) which measures temperature only.
Among these, two studies calibrated the Thermochron using a constant
water bath prior to experimentation and reported its validity [1,8].
Both studies showed linear correlations between Thermochrons and the
calibrated thermometer with correlation coecients larger than 0.99
and mean biases of + 0.121 C [1] and 0.09 C [8], which are better
lowing the application of individual correction formula to each Ther-
mochron, the accuracy was enhanced, as the mean bias was closer to
zero in both studies [1,8]. Harper Smith and his colleagues [1] con-
cluded that the high level of accuracy of the Thermochron implied that
if a 0.1 C dierence between the Thermochron and thermometer
were acceptable, then the time consuming calibration process would
not be necessary.
The Hygrochron is also a small and self-sucient device similar in
shape to the Thermochron, but measures both temperature and RH. It
can be set to either an 8-bit (0.5 C and 0.6% RH) or 11-bit (0.0625 C
and 0.04% RH) resolution. Operating ranges are 20 to 85 C and 0 to
100% for temperature and RH, respectively [13]. The Hygrochron has
been used in exercise testing [2], monitoring thermal conditions in an
operating theatre [14] and in sleep research [15,16]. Two studies have
reported the use of the Hygrochron for microclimate temperature and
RH measurements [14,16] and another two recorded skin temperatures
[2,16]. To our knowledge, no studies have conducted calibration or

Corresponding author.
E-mail address: mirim.shin@sydney.edu.au (M. Shin).

http://dx.doi.org/10.1016/j.physbeh.2017.06.019
Received 3 March 2017; Received in revised form 22 June 2017; Accepted 26 June 2017
Available online 27 June 2017
0031-9384/ 2017 Elsevier Inc. All rights reserved.
M. Shin et al. Physiology & Behavior 179 (2017) 208212

validation of RH.
Accurate and practical RH measurement is demanded in studies of
thermal comfort. Thermal discomfort is signicantly inuenced by
near-to-skin temperature or RH outside of the thermal comfort zone
[17,18]. The evaporation of sweat occurs when a gradient for RH exists
between the skin and the ambient surroundings allowing heat dissipa-
tion with skin cooling [19]. Therefore, this study aimed to validate the
RH measurement by the Hygrochron. This study assessed the Hygro-
chron sensor surface for RH detection; performed the calibration pro-
cess using a Humidity Calibration chamber; and reported the validity
and reproducibility of Hygrochron for the RH measurement. According
to Maxim Integrated, the manufacturer, the temperature sensor of Hy-
grochron is the same one used in the Thermochron. Two studies have
reported its accuracy to be within 0.1 C [1,8]. Nevertheless, we
have calibrated the temperature of the Hygrochron since RH is a tem- Fig. 2. Mean relative humidity measured by Hygrochrons for each sensor surface.
perature-dependent variable [20]. The temperature calibration data Sensor up: sensor facing-up away from the paper towel. Sensor down: sensor facing-down
towards the surface of paper towel; arrow bars with * indicate the p-value < 0.05 for the
pertinent to RH measurement only were reported.
specied time period.

2. Part 1: sensor surface for RH detection

According to the manufacturer, RH is measured via the surface of
the Hygrochron, which has a tiny opening on the protruding surface
that allows water vapour to pass through and reach the internal hu-
midity sensor (Fig. 1A). To evaluate the dierences in RH detected
between the two surfaces of the Hygrochron, four Hygrochrons were
placed on a wet paper towel, with two sensors facing up and the other
two facing down. The wet paper towel was chosen to create dierent
RH on either side of the sensor. Since RH (the ratio of the amount of
water vapour in the air required for saturation at a given temperature
and pressure) measurement was of interest, each Hygrochron was
placed in a designed holder to avoid direct contact between the sensor
surface and wet paper towel (Fig. 1B). The holder which was printed in
acrylonitrile butadiene styrene on a 3-D printer, holds the Hygrochron
2 mm above the wet paper towel and the holder's diameter was 16 mm
and height was 7 mm.
2.1. Methods
Four Hygrochrons were congured at a sampling rate of 5 s and at
11-bit resolution (0.0625 C/0.04%RH). They were rst placed on the
bench top at the laboratory for 5 min before being moved onto the wet
paper towel for the next 20 min: two Hygrochrons with the humidity
sensor facing up and the other two with the sensor facing down. The
ambient temperature and RH at the laboratory were 22.5 C and 50%,
respectively. Data were retrieved through the eTemperature software
(OnSolution Pty Ltd., Australia) and exported to Excel. t-Tests (SPSS
v22, Chicago, IL) were run to compare RH readings between sensors
facing up and down across the experimental time.
2.2. Results and discussion
There was no dierence in RH measured between sensor positioning
when all Hygrochrons were placed on the bench top. When the
Hygrochrons were on the wet paper towel, the RH measured by the
sensor facing down was signicantly higher than that facing up from
390 s onwards (p < 0.05). The maximum dierence was 31.2% RH
(Fig. 2). The high humidity recorded was expected, since the sensor
faced down on the wet paper towel. The ndings indicate that the
protruding surface of the Hygrochron (Fig. 1A) should be placed in the
direction of interest, e.g., a skin surface or the air in a humidity climate
chamber.
3. Part 2: calibration and validation of Hygrochron for RH
measurement
The calibration RH range applied was from 40 to 90% at a tem-
perature ranged from 32 to 37 C. These values were chosen to prepare
for a research study on thermal comfort and sleep that examined the
microclimate temperatures and RH between the skin surface and
sleepwear during sleep at 30 C and 50% RH. In general, an accuracy
of < 2% is considered acceptable for RH measurements [21], although
an accuracy of 5% at one temperature only (25 C) was reported for the
Hygrochron by the manufacturer. Thus, there is a need to assess the
accuracy of RH measurements at other temperatures.
3.1. Methods
The calibration of the Hygrochron was performed using a humidity
calibration chamber (Flow Divider Humidity Generator, FDC1-1C-1,
Tecnequip Enterprises Pty Ltd., Australia) to establish the validity of the
Hygrochron. The calibration chamber is able to generate humidity in
1% RH steps at an accuracy of 0.2% RH, ranging from 1 to 99% RH
and at any temperature from 1 to 99 C. The RH was generated by
mixing wet and dry air through two ow nozzles and circulating the
resulting mixed air using a motorized fan. A reference RH was obtained
by mixing accurately proportioned quantities of dry air at 0% RH and
saturated air at 100%. The chamber was positioned centrally within a

Fig. 1. A. Illustration of the Hygrochron data logger (http://www.elin.ru/iBDL/?topic=case); B. A Hygrochron in a designed holder.

209
M. Shin et al.
Fig. 3. Raw temperature and RH data of two Hygrochrons during the calibration pro-
cedure at a chamber temperature of 35.2 C.
20 l Perspex container lled with water. The water in the container was
heated or cooled by a temperature controlled heated water bath. Both
dry air and the saturated air are temperature conditioned before being
mixed in the main chamber. The chamber temperature was regularly
monitored using a calibrated platinum resistance thermometer (Leeds
and Northup model 8926).
The RH was calibrated for the range of 4090% at six temperature
levels from 32-37C. At each temperature level, the RH was increased by
10% every 15 min. This period was adequate to allow the RH to reach a
steady state (plateau) (see Fig. 3). The RH calibration process for each
temperature took approximately 2 h. Twenty-eight Hygrochrons were
congured at the sampling rate of 5 s and at 11-bit resolution
(0.0625 C/0.04%RH) and synchronised for the start of recordings. All
Hygrochrons were suspended in the calibration chamber by attaching
Hygrochrons on an adhesive tape. Once the calibration procedure was
completed for each temperature, the Hygrochron data were retrieved
through the eTemprature software and exported to the Excel for ana-
lysis. All 28 Hygrochrons except one reached the equilibrium plateau
for each RH increment for all temperature conditions as shown in
Fig. 3A, upper panel. The outlier did not reach a plateau when the
calibration chamber RH was set at 70 to 90% and this instability at high
RH was observed under all temperature conditions (Fig. 3B). Thus this
Hygrochron was excluded from data analysis.
The calibration of RH commenced at 37.50 min once the chamber
temperature reached a steady state. A: a typical behaviour of a
Hygrochron and B: a non-typical behaviour where a steady state was
not reached at RH of 7090%.
At each temperature condition (32 to 37 C), the mean temperature
measured by each Hygrochron was obtained by averaging the tem-
perature readings at a steady state over an approximately 10 min in-
terval. Similarly, the mean RH value for each RH increment (40 to 90%)
was obtained by averaging the RH readings at steady state over an
approximately 5 min interval (Fig. 3A). The mean temperature and RH
data for each Hygrochron were plotted against the mean temperature of
the calibrated thermometer and the mean chamber RH, respectively. A
linear regression line was then constructed for both temperature and
RH for each Hygrochron (Fig. 4). The linear regression equations were
used for the correction of measured temperature and RH. Following the
calibration process, six RH correction equations were generated for
210
Physiology & Behavior 179 (2017) 208212
each Hygrochron at each temperature. The paired t-test was applied to
compare the raw and corrected RH against the chamber RH (SPSS v22).
Data for temperature calibration were not reported in this trial as
previous studies found that mean bias was approximate 1 C [1,8] and
little eect on RH was observed for changing temperatures within the
32 C to 37 C range of interest.
3.2. Results and discussion
We veried that a tight correlation existed between the temperature
measured by the Hygrochron iButton and the calibrated thermometer.
The Pearson correlation coecients for all Hygrochrons were > 0.99.
The slope coecients averaged 0.92 0.03 C (range = 0.86 to
1.01 C) and the intercepts averaged 2.25 1.14 C (range = 0.48
to 4.53 C). The linear regressions between RH of the Hygrochrons and
the calibration chamber were also highly signicant with all correlation
coecients > 0.99. The slope calibration coecients averaged
1.16 0.03% (range = 1.08 to 1.23%), and the intercepts ranged
11.37 2.12% (range = 16.5 to 5.95%).
The Hygrochron overestimated RH (range 1.82 to 4.85%) at the
lower RH range (4060% RH of chamber readings) but underestimated
RH (range 0.78 to 3.16%) at the higher RH range (8090% RH), at all
temperature conditions (upper panel, Table 1). Hygrochron RH was
signicantly dierent to the chamber RH (mean dier-
ence = 1.08 2.55%, p < 0.001). After applying the correction
equation for each Hygrochron, there was no signicant dierence be-
tween Hygrochron RH and the chamber RH (mean dier-
ence = 0.002 0.82%, p = 0.9) (lower panel, Table 1).
The measured Hygrochron RH deviated from the chamber RH by up
to 4.85% at the lower chamber RH of 4060% and 3.16% at the higher
chamber RH of 90% (upper panel, Table 1). In addition, of the 28
Hygrochrons, one did not equilibrate at RH level starting from 70% (see
Fig. 3B). These ndings suggest that Hygrochrons should be calibrated
prior to use in a research setting. A limitation of this study was that the
response time of Hygrochron could not be evaluated, since the chamber
RH could not be recorded continuously. Notably, the manufacturer
reported that the humidity response time constant was 30 s. In addition,
the RH calibration only reached 90% due to the RH generation capacity
of the calibration chamber, which ranges from 1 to 99%, where we
employed a 10% step increment.
4. Part 3: evaluation of reproducibility of RH measurement by
Hygrochron
In addition to validity, the reproducibility of measurements is es-
sential to produce consistent results under identical conditions using
the same instrument and methods. The same sets of Hygrochrons were
tested twice under the same condition over a three month interval to
determine the repeatability of RH measurements.
4.1. Methods
Eleven Hygrochrons (congured at a sampling rate of 10 s with the
high resolution of 11-bit) were suspended in the humidity calibration
chamber. The calibration chamber RH was increased by 10% from 50 to
90% every 15 min while the temperature of the chamber was main-
tained at 35.0 C. Three months later the same Hygrochrons (congured
at a sampling rate of 5 s with the high resolution of 11-bit) were put
through the same process at 35.2 C. It should be noted that dierent
sample rates were set for the Hygrochrons in trial 1 and 2 according to
experimental durations. The RH values were corrected using individual
correction equations. The mean RH value for each RH increment (50 to
90%) for the 35.0 C and 35.2 C trials were compared using an intra-
class correlation (ICC) (two-way, mixed absolute agreement and single-
measures) to assess the degree of consistency and agreement. Paired t-
tests were also applied to compare the two data sets (SPSS v22).
M. Shin et al. Physiology & Behavior 179 (2017) 208212

Fig. 4. Linear regression of Hygrochron temperature and thermometer temperature in the chamber (left) and Hygrochron RH and the chamber RH (right).

4.2. Results and discussion Table 2

Mean Hygrochron RH at each chamber RH at two time points over a three month interval,
Table 2 shows the mean Hygrochron RH at each chamber RH at the and their dierence in RH (all values are presented as mean standard deviation).

initial and three-month time-point. Overall, the RH data collected at the Trial 1 Trial 2 Dierence between trials
two time-points were highly reproducible, as reected by the ICC range
of between 0.994 and 1.000, although the greatest reproducibility 50% 0.79 0.95 0.29 1.02 0.50 0.95
(lowest dierence) was observed at 80% RH (Table 2). In addition, 60% 0.98 0.81 0.38 1.01 0.60 1.14
70% 1.08 0.66 0.48 0.34 0.59 0.73
there was no signicant dierence between the two data sets with a 80% 0.17 0.74 0.15 0.10 0.02 0.80
mean dierence of 0.202 1.05% (p = 0.159). 90% 1.11 1.21 0.44 0.40 0.67 1.12

5. Conclusion
defect. There are many possible applications with the use of the
This study tested and detected the sensor surface of the Hygrochron Hygrochron. Future studies could use the Hygrochron, held in a holder,
iButton, reported the calibration and validation process, and the re- for the measurement of skin-clothing microclimates.
producibility of RH measurements. Following calibration, the improved
accuracy was 0.002 0.82%. Reproducibility measures showed
high correlations (ICC range of between 0.994 and 1.000). The Acknowledgments
Hygrochron is valid and reproducible for RH measurements when
placing the protruding surface in the direction of interest (skin surface The authors thank Australian Wool Innovation Ltd. (AWI) (OF-
or the air of a chamber). However, it should be emphasized that each 00213) for nancial support for the project that yielded material dis-
Hygrochron should be calibrated at the desired RH and temperature cussed in this article.
prior to use to improve its accuracy and to detect any manufacturing

Table 1
Mean Hygrochron RH at each chamber temperature and RH before and after correction (the data are shown as mean standard deviation (condence interval)).

32.3 C 33.2 C 34.7 C 35.2 C 36.2 C 37.1 C

Hygrochron RH (before correction)

40% 44.85 0.94 44.43 0.77 44.17 0.72 43.41 0.74 43.60 0.80 43.90 0.80
(44.4845.22) (44.1244.73) (43.8844.45) (43.1243.71) (43.2843.91) (43.5944.22)
50% 53.26 0.84 53.20 1.21 52.96 1.43 52.94 1.33 52.71 1.18 53.28 1.26
(52.9353.59) (52.7253.68) (52.3953.53) (52.4153.46) (52.2453.18) (52.7853.78)
60% 62.66 1.01 62.58 1.28 62.33 1.02 61.84 1.23 61.82 1.16 62.09 1.31
(62.2663.06) (62.0863.09) (61.9362.74) (61.3562.32) (61.3662.28) (61.5862.61)
70% 70.79 0.85 71.08 0.86 70.85 0.69 70.82 0.88 70.76 0.79 70.97 0.85
(70.4571.12) (70.7471.43) (70.5871.12) (70.4771.17) (70.4471.07) (70.6371.3)
80% 78.96 0.81 79.12 0.79 78.98 0.73 79.15 0.88 79.07 0.78 79.22 0.77
(78.6479.28) (78.8179.43) (78.6979.27) (78.8079.50) (78.7679.37) (78.9179.52)
90% 86.98 0.85 86.84 0.69 86.92 0.87 87.73 0.87 87.56 0.77 87.25 0.83
(86.6487.31) (86.5687.11) (86.5787.26) (87.3888.07) (87.2587.86) (86.9287.58)

Hygrochron RH (after correction)

40% 39.71 0.20 39.57 0.33 39.71 0.34 39.51 0.21 39.66 0.18 39.44 0.20
(39.6339.79) (39.4439.70) (39.5739.84) (39.4239.58) (39.5939.73) (39.3639.52)
50% 49.65 0.56 49.85 1.19 49.86 1.25 50.27 1.08 50.01 0.92 50.23. 1.20
(49.4349.88) (49.3850.32) (49.3650.35) (49.8450.70) (49.6550.37) (49.7650.71)
60% 60.76 0.43 60.81 1.18 60.67 1.22 60.31 1.10 60.34 1.04 60.36 1.16
(60.5960.93) (60.3461.28) (60.1961.15) (59.8860.75) (59.9360.75) (59.9060.82)
70% 70.36 0.21 70.77 0.53 70.50 0.66 70.46 0.34 70.49 0.30 70.58 0.32
(70.2870.44) (70.5670.98) (70.2470.76) (70.3370.60) (70.3770.61) (70.4570.71)
80% 80.02 0.16 80.18 0.44 79.89 0.60 79.88 0.08 79.93 0.09 80.07 0.09
(79.9680.08) (80.0080.35) (79.6580.13) (79.8479.91) (79.8979.96) (80.0380.11)
90% 89.50 0.19 89.21 0.44 89.05 0.76 89.57 0.43 89.57 0.42 89.32 0.41
(89.4289.57) (89.0489.39) (88.7589.36) (89.4089.74) (89.4089.74) (89.1689.48)

211
M. Shin et al. Physiology & Behavior 179 (2017) 208212

References
[11]
[1] A.D. Harper Smith, D.R. Crabtree, J.L.J. Bilzon, N.P. Walsh, The validity of wireless
iButtons and thermistors for human skin temperature measurement, Physiol. Meas.
31 (2010) 95114. [12]
[2] Y. Zhai, C. Elsworth, E. Arens, H. Zhang, Y. Zhang, L. Zhao, Using air movement for
comfort during moderate exercise, Build. Environ. 94 (2015) 344352.
[3] A. Venable, R. Williams, B. McFarlin, Gender dierences in skin and core body [13]
temperature during exercise in a hot, humid environment, Int. J. Exerc. Sci. (2013).
[4] R. Areas, L. Duarte, L. Menna-Barreto, Comparative analysis of rhythmic parameters [14]
of the body temperature in humans measured with thermistors and digital ther-
mometers, Biol. Rhythm. Res. 37 (2006) 419424. [15]
[5] M. Zornoza-Moreno, S. Fuentes-Hernandez, M. Sanchez-Solis, M.A. Rol, E. Larque,
J.A. Madrid, Assessment of circadian rhythms of both skin temperature and motor
activity in infants during the rst 6 months of life, Chronobiol. Int. 28 (2011) [16]
330337.
[6] J. Sarabia, M. Rol, P. Mendiola, J. Madrid, Circadian rhythm of wrist temperature in
normal-living subjects: a candidate of new index of the circadian system, Physiol. [17]
Behav. 95 (2008) 570580.
[7] M.J. Hasselberg, J. McMahon, K. Parker, The validity, reliability, and utility of the
iButton for measurement of body temperature circadian rhythms in sleep/wake [18]
research, Sleep Med. 14 (2013) 511.
[8] W.D. van Marken Lichtenbelt, H.A. Daanen, L. Wouters, R. Fronczek, R.J. Raymann, [19]
N.M. Severens, et al., Evaluation of wireless determination of skin temperature
using iButtons, Physiol. Behav. 88 (2006) 489497. [20]
[9] Y. Inoko, K. Yoshimura, C. Kato, O. Morita, M. Kohno, Ecacy and safety of tem-
perature data loggers in measuring compliance with the use of oral appliances, [21]
Sleep Biol. Rhythm. 7 (2009) 188192.
[10] G.P. Nicholson, M.W. Ferguson-Pell, K. Smith, M. Edgar, T. Morley, The objective
measurement of spinal orthosis use for the treatment of adolescent idiopathic sco-
liosis, Spine 28 (2003) 22432250.
B.M. Benish, K.J. Smith, M.H. Schwartz, Validation of a miniature thermochron for
monitoring thoracolumbosacral orthosis wear time, Spine 37 (2012) 309315
(03622436).
Y. Koike, H. Sano, T. Kinjyo, I. Imamura, O. Masahiro, M. Goto, et al., Shoulder
surface temperature and bone scintigraphy ndings in patients with rotator cu
tears, Ups. J. Med. Sci. 116 (2011) 142147.
Maxim, iButton Hygrochron Temperature/Humidity Logger with 8KB Data-Log
Memory. Maxim Integrated, (2016).
S. Mazzacane, C. Giaconia, S. Costanzo, A. Cusumano, A survey on the thermal
conditions experienced by a surgical team, Indoor Built Environ. 16 (2006) 99109.
M. Shin, M. Halaki, P. Swan, A. Ireland, C.M. Chow, The eects of fabric for
sleepwear and bedding on sleep at ambient temperatures of 17 C and 22 C, Nature
Sci. Sleep 8 (2016) 121131.
C. Song, Y. Liu, X. Zhou, X. Wang, J. Li, J. Lui, Temperature eld of bed climate and
thermal comfort assessment based on local thermal sensations, Build. Environ. 95
(2016) 381390.
A.P. Gagge, J. Stolwijk, J. Hardy, Comfort and thermal sensations and associated
physiological responses at various ambient temperatures, Environ. Res. 1 (1967)
120.
G. Havenith, Interaction of clothing and thermoregulation, Exog. Dermatol. 1
(2003) 221230.
K. Okamoto-Mizuno, K. Mizuno, Eects of thermal environment on sleep and cir-
cadian rhythm, J. Physiol. Anthropol. 31 (2012).
R. Schellenberg, The trouble with humidity: the hidden challenge of RH calibration,
Cal Lab. (2002) 4042.
J.F. Thompson, Psychrometrics and perishable commodities, in: A.A. Kader (Ed.),
Postharvest Technology of Horticultural Cropts, 2002, pp. 129134.

212