Data were gathered from 38 male participants (age = 36 ± 3 years) on taste characteristics of a range of l-menthol concentrations in a randomised, double-blind, crossover survey. In subsequent experimental trials, 31 healthy non-acclimated males (age = 21 ± 2 years; body mass = 79.1 ± 9.2 kg; stature = 180.3 ± 6.7 cm) were recruited to participate in two independent randomised, double-blind, crossover experimental studies, exploring the dose–response eff1ect of l-menthol mouth rinse on exercise performance (study 2, n = 16) or, in a follow-up study, administration of increasing or decreasing doses of l-menthol on exercise performance (Study 3, n = 15). Randomisation was conducted by generating random numbers for each condition for all participants using online software (Urbaniak and Plous 2015) and blinding was performed by a person that was not on the experimental research team and all solutions were administered with random letters. Participants were blinded to the original hypothesis of the study and informed that the effect of differing mouth rinses on exercise in the heat was being investigated. Participants abstained from alcohol, caffeine, and strenuous exercise in the 24 h leading up to the day of testing. None of the participants had visited a hot country in the previous three months and all testing was conducted during the UK winter months of November to April. Participants visited the laboratory on between 5 and 6 separate occasions, each separated by at least 72-h. A-priori sample size was calculated using G*Power (version 3.1.9.6). Given the effect size (ηp2 = 0.896; (Flood et al. 2017)), we reported previously for differences in power output using an RPE-16 protocol with l-menthol, a sample size of 10 was deemed sufficient to identify differences between groups with a statistical power of 0.95. More participants were recruited to account for experimental attrition. Informed written consent was obtained from each participant before commencing the study. Ethical approval was provided by the Newcastle University ethics committee.

In study 1, participants were asked to orally rinse a range of l-menthol concentrations (0.064 mM = 0.001% to 64 mM = 1%) delivered at a temperature of ~ 31 °C, equivalent to oral temperature neutrality (Green 1985). Laboratory conditions were maintained at 20 ± 0.6 °C and testing was controlled at the same time of day ± 2 h. Participants rinsed 25 ml solutions for 10 s before expectorating into a bucket, they were told to keep their mouth closed to prevent evaporative cooling. At 30 s, participants were asked to rate the intensity of coolness, irritation and pleasantness in the mouth. Ratings were made using the oral labelled magnitude scale, adapted from (Cliff and Green 1994a, b; Green et al. 1993). Magnitudes of somaesthetic descriptors were: Barely detectable = 1, weak = 6, moderate = 16, strong = 34, very strong = 50, strongest imaginable = 95 (Green et al. 1993). Data were then presented according to mean intensities. Different formulations were tested at least > 24 h apart to reduce the potential for oral desensitisation.

Study 2 and 3 were comprised of independent participant groups, however, the familiarisation and exercise protocols were the same. Participants were thoroughly briefed on the RPE protocol, as we have previously described (Flood et al. 2017). Before commencing the study, participants conducted a number of RPE-clamp familiarisation tasks to reduced variability. Firstly, participants undertook an RPE ramp test to identify the exercise intensity at each stage associated with the RPE scale. This task was used to calibrate the individual’s understanding of their own RPE. Following a 5‐min rest period, a series of confirmation trials were conducted, starting at 120 W and controlling power output until they achieved an effort that they equated to an RPE-16, across a 3 min period. These trials were continued until the individual could reliably determine the exercise intensity at an RPE-16 within ± 10 W, this took on average 3 attempts. Partial familiarisation of the full experimental RPE-clamp test protocol was then conducted in an environmental heat chamber to experience heat stress and to reduce a subsequent learning effect. Participants were also given significant time to discuss and understand the RPE protocol with the researchers both before and after this initial familiarisation performance trial. Subsequent randomised experimental trials were separated by at least > 72 h and conducted in an environmentally controlled heat chamber, at a temperature of 35.0 ± 0.8 °C and relative humidity 30 ± 3.3%. For each participant, the experimental trials were conducted at the same time of day (± 2 h) to eliminate the effect of circadian variation. All exercise was conducted on an electronically-braked cycle ergometer (Lode Excalibur Sport, The Netherlands).

Upon entering the heat chamber, the participants conducted two standardised RPE warm-up procedures, as used during familiarisation. Participants were then instructed to cycle at a power output that was perceived to represent an RPE of 16 on the 6-to-20 Borg scale (Borg 1982) and to adjust their power output such that an RPE of 16 was maintained. An RPE of 16 represents a verbal cue of between ‘hard’ and ‘very hard’ on the Borg Scale. Participants completed the fixed-RPE protocol and were administered an l-menthol or control mouth rinse immediately prior to and throughout (at 10 min intervals) the experimental protocol. The highest average 30-s power output achieved during the first 3-min of the fixed RPE trial was recorded and participants exercised until their power output declined to 70% of this initial value. The trial was stopped when power output fell below this value for 2-min. Standardised feedback every ~ 2 min was given to remind participants to maintain an RPE of 16. Participants were encouraged to constantly reassess whether they were still exercising at RPE-16. Participants were blinded to distance covered, elapsed time, heart rate, power output.

Euhydration was established prior to every trial by identifying urine osmolality < 715 mOsm/Kg H2O (Shirreffs and Maughan 1998) (Pocket Osmochek, Vitech Scientific Ltd, West Sussex, UK) average hydration was 435 ± 188 mOsmols/kg H2O. Participants recorded semi-nude body mass (cycling shorts only) prior to entering the heat chamber and immediately following the completion of the experimental trial after wiping off sweat with a towel. No water was ingested during exercise in the heat. Body core temperature was reported using a self-inserted temperature probe (PROACT Medical Ltd. PRTP11112, Northamptonshire, UK) 10 cm past the anal sphincter and logged continuously on a squirrel data logger (Squirrel SQ2040 2F16 data logger, Grant Instruments, Cambridge, UK). Heart rate was recorded continuously throughout the trials (Polar Heart Rate Monitor M400, Warwick, UK).

Participants were thoroughly briefed on the RPE scale before commencing the fixed RPE trials. In addition, participants where familiarised with the thermal sensation (TS) scale which was recorded on an adapted ASHRAE 9-point analogue sensation scale where − 4 = “very cold”, 0 = “neutral”, and 4 = “very hot” (Zhang et al. 2004). Scales were laminated and held in front of the participants to physically indicate scores whilst exercising. Subjective ratings of TS were recorded in 1.0 increments every 10 min during the experimental trials.

Each rinse consisted of a 25 ml colourless solution. The first was given in the 30 s prior to the start of the trial and then at 10-min intervals thereafter. Participants were instructed to swill the solution in their mouth for 10 s before spitting it into a bowl. Five different l-menthol solutions were formulated at a range of concentrations: 0.001%, 0.01%, 0.1%, 0.5% and 1% and compared to a non-calorific flavoured control. In study 2, l-menthol was delivered at each interval at the desired concentration (between 0.001 and 0.5%). In study 3, two differing protocols were used. The descending protocol: 0.5% followed by 0.1% and 0.01% and then 0.01% thereafter until test termination. The ascending protocol: 0.01% followed by 0.1% and 0.5% and then 0.5% thereafter until test termination. The constant protocol: 0.01% delivered throughout (Fig. 1). l-Menthol solutions were made using menthol crystals (≥ 99% food grade l-menthol, Sigma-Aldrich, UK) and first dissolved in mono-propylene glycol (Special Ingredients Ltd, Chesterfield, UK) to facilitate solubilisation and then serial diluted in de-ionized water. To avoid recrystallization of menthol at the highest concentrations, solutions were prepared on a daily basis. A control mouth rinse was formulated using a strawberry flavoured non-calorific artificial sweetener (FlavDrops, MyProtein, Norwich, UK) and mono-propylene glycol, which dissolved in de-ionized water. Solutions were then stored at 5 °C and prior to use warmed back to ~ 31 °C passively in the environmental chamber prior to administration.

Schematic illustrating the concentration application protocol in study 2 A over a range of l-menthol concentrations from 0.001 to 0.5%. In study 3 B an ascending protocol used low to high concentrations (0.01–0.5%), C a descending protocol used high to low concentrations (0.5–0.01%), D a constant protocol applied the standard reported concentration of 0.01% at regular intervals and E a control protocol where a strawberry flavoured non-calorific artificial sweetener was applied at regular intervals

All statistical analyses were performed using SPSS (IBM SPSS statistics 22 Inc, USA). A two-way analysis of variance (ANOVA) for repeated measures was used to test for within-group effects across time in all conditions. If sphericity was violated a Greenhouse–Geisser correction was applied. When a significant difference was found for main effect (trial or time), post hoc pair-wise comparisons were made incorporating a Bonferroni adjustment. Magnitude of effect was calculated with partial eta-squared (ηp2) according to the following criteria: 0.02, a small difference; 0. 13, a moderate difference; 0.26 a large difference (Cohen 1988). Differing trial durations meant that power data were normalised with respect to time, and for illustration purposes with respect to starting power output. Perceptual data using the oral labelled magnitude scale was analysed using the non-parametric Friedman test. Data are presented as mean ± SD, significance was set at P ≤ 0.05.