In this study we reported the most up-to-date firefighter hand anthropometry through a national firefighter anthropometric study and evaluated the current use and fit of structural firefighting gloves to the firefighter population. A new perspective for enhancing glove sizing scheme was proposed accordingly. Four hypotheses were tested: (a) Firefighter hand dimensions were different among gender, race/ethnicity, and age groups; (b) current firefighter hand dimension ranges are greater than the specifications outlined in the NFPA 1971 glove-sizing scheme; (c) there is no association between choosing an NFPA-matched size and the fit of the gloves reported by firefighters; and (d) glove fit is multidimensional and is associated with hand shape (hand and finger breadth-to-length contrast) aside from the overall hand size.

Fourteen dimensions relevant to the design of gloves were measured in this study. They were hand length, hand breadth, palm length, palm breadth, thumb and finger lengths, and thumb and finger breadths. The graphical descriptions and definitions of these dimensions are presented in Figure 1 and Appendix A, respectively. The NFPA 1971 standard on gloves contains additional information on hand and finger circumferences; this present study did not collect these data. The Hand Anthropometry of U.S. Army Personnel data provided two equations for converting the NFPA hand circumference ranges to hand/palm breadth ranges, which were employed in this study to evaluate the NFPA sizing system. The equations are Hand/Palm Breadth (males) = 4.93 + 0.4 × Hand Circumference (coefficient of determination r² = .93; Greiner, 1991, p. 432) and Hand/Palm Breadth (females) = 0.62 + 0.42 × Hand Circumference (r² = .937; Greiner, 1991, p. 434).

A stratified sample of 863 male firefighters (three age by three race/ethnicity combinations) and 88 female firefighters (three age groups) for a total of 951 firefighters participated in the study. Their right-hand scans were recorded using a flatbed scanner (Epson Perfection 1260; 10,200 × 14,040 pixels at 1,200 dpi) as part of a national firefighter anthropometry survey (Hsiao et al., 2014).

The sampling plan was based on a statistical power estimation for an assessment of differences in anthropometry among gender, race, and age groups who represented the 1,136,650 firefighters documented in the U.S. Fire Department Profile Through 2005 (Karter, 2006). Data collection was conducted in four geographic locations (Rockville, Maryland; Phoenix, Arizona; Philadelphia, Pennsylvania; and Fort Worth, Texas; Table 1). The number of participants in each region was assigned based on the size of the population in that region in the 2000 U.S. Census (U.S. Census Bureau, 2001a, 2001b, 2001c) with an assumption that the number of firefighters was proportional to the size of the population they serve. The detailed sampling plan, accounting for geographic density of racial/ethnic distributions, was presented in Hsiao et al. (2014). These firefighter profiles represented the best available and most up-to-date firefighter population distribution data at the study-planning stage in 2007. An oversampling of female firefighters (88/951 = 9.3%) was necessary, in lieu of 4.2% per the Household Data Survey (U.S. Department of Labor, 2006), to address some issues (i.e., hand and finger shape associated with glove design) that were critical to females.

The data collection stations consisted of a briefing table, a changing area, and a space with sufficient lighting for measurements. Participants were approached through leaders of regional fire stations at four data collection sites. Data collection was performed during 2009 to 2012. Upon arrival at the NIOSH field laboratory that was established at a fire station, the firefighters were given a brief overview of the study. Participants then signed a consent form and completed a questionnaire related to experience with fire apparatus and protective gear. As part of the National Firefighter Anthropometry Survey (Hsiao et al., 2014), a two-dimensional (2-D) hand scan of the right hand was recorded for each participant. Fourteen dimensions were then extracted using image digitizing software CorelDraw. Participants’ biographical information (e.g., gender, race/ethnicity, and age), the size of gloves used, and the self-assessment of glove fit (too tight, acceptable, or too loose) were also logged. The self-assessment of glove fit was based on the gloves that the participant has used and brought to the study. A “good fit” of a glove was defined as that the glove was not too tight or too loose at any finger and hand area for which a firefighter performed tasks during regular operations.

Literature has shown no significant difference in finger and hand length measurements between using the 2-D scanning method and the traditional caliper-based measurement method; additionally, the scanning method took considerably less time than the traditional measurement method (Yu, Yick, Ng, & Yip, 2013, p. 389). Hand/palm breadth obtained from image digitization in a hand abduction posture (fingers open), on average, is 5 mm greater for men and 3.6 mm greater for women than that of a traditional measurement in a fingers-closed pose (Greiner, 1991).

Statistical analyses on the 14 hand measurements were performed for their arithmetic means and standard deviations (Tables 2 through 4). A previous publication has verified that the raw data from the NIOSH National Firefighter Anthropometry Survey can be employed directly into protective gear design applications without major weighting or abnormality concerns (Hsiao et al., 2014). Data from eight participants were excluded in the analyses because three of them have a partial amputation on one of their fingers and five participants’ hand scans were missing due to a scanner malfunction. Analysis results show significant differences in means of the dimensions by gender (14 dimensions), race/ethnicity (14 dimensions), and age group (all except four finger lengths).

The NFPA standard on gloves categorizes glove sizes mainly by hand length and circumference. In this current study, hand length and palm breadth from the 14 collected hand dimensions were selected for the sizing coverage analysis because they were compatible with the dimensions used in the NFPA standard (Table 5). The hand length is defined as the combined palm length and middle finger length. The hand circumference range for each glove size outlined in the NFPA 1971 standard was converted to a palm breadth range, using Hand/Palm Breadth (mm; males) = 4.93 + 0.4 × Hand Circumference as the upper boundary and Hand/Palm Breadth (mm; females) = 0.62 + 0.42 × Hand Circumference as the lower boundary (Greiner, 1991) as reported in the Method section. In this study, both palm breadth and hand breadth were collected with fingers open naturally. These two measurements are highly correlated (r² = .93). The hand breadth data were slightly affected by the level of opening of fingers. Palm breadth measurements were less affected and thus were used without adjustment in this assessment.

A comparison of the hand length and palm breadth data of the sampled firefighter population against those of the NFPA 1971 glove sizing scheme revealed that 72% (617/855) of men and 96% (84/88) of women are covered by the current NFPA 1971 glove-sizing chart (Table 5). Women had a higher coverage rate than men, χ² = 22.69 > χ²(1, .05) = 3.840, p < .001. The hand dimension ranges outlined in the NFPA 1971 glove standard in general underrepresented national firefighter hand size variation (Figure 2a). An adjustment of the palm breadth data from fingers-open posture to fingers-closed pose showed that 91% (777/855) of men and 97% (85/88) of women are covered under the current NFPA 1971 glove-sizing map. The coverage rates were not statistically different between genders, χ² = 3.32 < χ²(1, .05) = 3.840, p = .07. The NFPA sizing scheme still understated hand size distribution of the firefighter population (Figure 2b).

Of the 943 participants who had the complete hand dimension measuments, 586 reported the glove sizes they used. Of the 586 responses, 434 (74%) indicated that their gloves fit their hands well, whereas 152 (26%) reported that the gloves they used did not fit them adequately (Table 6). For men, there was 76% (415/547) fit and 24% (132/547) nonfit. For women, the ratio was 49% (19/39) fit and 51% (20/39) nonfit. The difference in glove-fit distribution is significant between men and women, χ² = 15.55 > χ²(1, .05) = 3.840, p < .001.

Also, of the 434 participants who had well-fitting gloves, 160 (37%) used glove sizes that matched the NFPA 1971 standard sizing scheme according to their hand dimensions (Table 6). In contrast, of the 152 participants who had poor-fitting gloves, 59 (39%) used glove sizes that matched the NFPA 1971 standard sizing scheme according to their hand length and palm breadth. No association was found between choosing an NFPA-matched size and the fit of the gloves, χ² = 0.18 < χ²(1, .05) = 3.840, p = .67.

While firefighter gloves are designed and used for operation with fingers open and closed, traditional hand measurement approaches and existing standards tended to report hand data in a fingers-closed posture (like mittens). Two equations were available for converting the raw data between the two postures: Hand/Palm Breadth Digitized (mm; fingers open; males) = 16.3 + 0.87 × Hand Breadth Measured (fingers closed; Greiner, 1991, page 279), and Hand/Palm Breadth Digitized (mm; fingers open; females) = 11.74 + 0.90 × Hand Breadth Measured (fingers closed; Greiner, 1991, p. 285). Using the converted data (fingers-closed posture), we again found no association between choosing an NFPA-matched size and the fit of the gloves, χ² = 1.28 < χ²(1, .05) = 3.840, p = .26 (Table 6).

A principal component analysis (PCA), based on 12 hand dimensions (lengths and breadths of palm, thumb, and fingers) was performed to explore the most distinguishable features of hand and fingers in defining hand sizes and shapes among firefighters to identify the direction for an improved glove-sizing plan (Table 7). Since multivariate normality assumptions are of secondary importance in this application (Jackson, 1997; Jolliffe, 2002), data of men and women were combined in this PCA analysis for its simplicity. Hand length and hand breadth were excluded in this PCA because of their high correlation with palm length and palm breadth, respectively (r = .93; see Hsiao et al., 2005, pp. 346–347). The result shows that the first two principal components are critical factors with eigenvalues of 6.74 and 2.42, respectively (Kaiser, 1960). They accounted for 56.2% and 20.1% of the total hand anthropometry variations (Table 7). The first principal component (PC1) represents the overall hand size (all dimensions as a group), and the second component (PC2) represents hand/finger breadth-to-length contrast as a whole.

Palm breadth and index finger length accounted for more variance of overall hand size (10.7% and 9.7%, respectively) than that of other dimensions. Middle finger breadth and index finger breadth accounted for more variance in hand shape (i.e., breadth-to-length contrast; 13.9% and 12.5%, respectively) than that of other dimensions. Using a weighted approach (i.e., 56.2% for PC1 and 20.1% for PC2) to calculate the influence of the 12 hand dimensions revealed that the middle finger length is the most contributory factor in defining hand sizes and shapes among firefighters (9.3% × 56.2 + 10.4% × 20.1 = 7.32; Table 7). Palm breadth, middle finger (length and breadth), and index finger (length and breadth) are the most distinguishable features of hand and fingers in defining hand sizes and shapes among firefighters and are therefore advantageous parameters for inclusion in future glove-sizing charts.

A further analysis of PC1 and PC2 indicated a significant mean difference of PC1 scores by gender (−1.59 for women and 0.16 for men, p < .001), and so did PC2 scores by gender (−0.35 for women and 0.04 for men, p < .001), which have practical implications in glove sizing and design. Men have a wider spread in hand/finger breadth-to-length contrast than women, and women may benefit from one or two gender-specific glove sizes for their smaller overall hand size and narrower hand/finger breadth-to-length contrast (i.e., elongated) as compared to those of men (Figure 3a). PC1 and PC2 are orthogonal and uncorrelated by definition when overall data are used (i.e., men and women are combined; Figure 3a). Scatter plots of PC1 against PC2 by gender show different trends between men and women for correlation between PC1 and PC2 (Figure 3b). PC1 and PC2 remain uncorrelated for men (correlation coefficient r = −.056); a negative correlation was observed for women (r = −.228, p < .05). Although this finding is likely due to the fact that men dominate the sample, so the two PC factors will be orthogonal to each other for them, the data trend revealed that women with a larger overall hand size tend to have smaller hand/finger breadth-to-length ratio.

In traditional protective gear fit research and design improvement, a practical and commonly used approach is to utilize the data from the groups who had good fit of their gear and apply the sizing strategy to the nonfit groups or the whole intended group. Given the significant mismatch between existing glove-sizing specifications and hand characteristics, such as the hand dimensions, user selection of glove size, and the existing glove-sizing specifications described in the current study, the same strategy would not yield a satisfactory result (Table 6). A revamping of the current sizing system by incorporating both an overall hand/fingers size and breadth-to-length contrast in the sizing scheme would be advisable. In light of the need for improved coverage (for men and women) and accommodation (for multiple finger dimensions), a cluster approach was employed. Clustering is a process that is used to group a set of objects in such a way that objects in the same group (a cluster) are more similar to each other than to those in other groups (clusters).

Although women have different distinct features in hand size and shape from men (Figure 3a), the naturally formed clusters of factor scores (i.e., PC1 and PC2 coordinates) of women pointed to the simplicity of an overall-group cluster analysis (Figure 4) over a combination of results from two gender-specific cluster analyses (Figure 3b). We propose a seven-size system to cover 97.5% of the sampled population. This system would not overburden the glove manufacturing industry in glove production or national firefighter stations in stocking gloves, while providing firefighters with a good chance of finding a well-fitting glove size. A cluster analysis was employed to identify the best clustering, in which the central size was predetermined for factors scores less than or equal to 1 (i.e., the distance from PC1 and PC2 coordinates to the center of PC1-PC2 plot is less than or equal to 1; Figure 4), and the remaining six sizes were defined through an iterative process where the sum of distances among the center of the six clusters was maximized.

There were two options to determine the six noncentral clusters. The first option was to use PC1 and PC2 scores. The second option was to cluster the 12 hand dimensions (lengths and breadths of palm, thumb, and fingers) that were used to derive PC1 and PC2. The first option has an advantage of having near-even clusters since only two variables (i.e., PC1 and PC2) were employed. The second option has a potential advantage of having a smaller standard deviation within each cluster as compared to the first option and thus was used in this paper. The final seven-cluster system contains Small, S-M, S-M Wide, Central/Medium, M-L Wide, M-L, and Large sizes (Figure 4) and the hand dimensions of the sizes (i.e., the clusters) are tabulated in Table 8.

This study reports significant differences in means of firefighter hand dimensions by gender, race/ethnicity, and age groups. The results on gender and race/ethnicity effects are generally in line with existing literature. Men, on average, have greater hand length and hand breadth than women (Gordon et al., 1989; Greiner, 1991; Pheasant & Haslegrave, 2006; Poston, 2000). Blacks have more elongated hands than other race/ethnicity groups (Courtney, 1984; Greiner, 1991; Martin et al., 1975; White, 1980). Information on age effect on hand dimensions of adults is lacking in the literature. This current study shows that the older firefighter group (≥40 years old) has larger hands than the younger group. The outcomes reaffirmed the importance of stratified sampling that adequately represents gender, race/ethnicity, and age distribution in the target population. It directly affects the level of accommodation of gloves to firefighters and glove design applications.

Several databases of hand anthropometry, mostly from military personnel, are available for exploring the differences in hand dimensions between firefighters and other occupational groups (Churchill, Churchill, McConville, & White, 1977; Clauser et al., 1972; Donelson & Gordon, 1996; Garrett, 1970a, 1970b; Gordon et al., 1989, Greiner, 1991; White & Churchill, 1971; Table 9). Male firefighters have larger mean hand length (196.6 mm) than Army and Marine Corps men (190.3–194.1 mm) but not Air Force men (197.2 mm). Female firefighters also have larger mean hand length (182.7 mm) than Army and Marine Corps women and Air Force nurses (177.4–180.7 mm) but not Air Force women as a group (183.8 mm). Literature has shown no difference in finger/hand length measurements between using 2-D flatbed scanning and traditional measurement methods (Yu et al., 2013). These hand dimension differences reflect occupational variances.

From the same surveys, firefighters have larger hand breadth (97.2 mm for men and 87.4 mm for women) than military personnel (89.0–90.4 mm for men and 75.5–79.5 mm for women). It is worth noting that a fingers-open hand posture was used for scanning in this study, and the hand breadth was extracted through a digitization process. Greiner (1991) reported an average 5-mm increase for men and 3.6-mm for women in hand breadth measurement when comparing a digitization approach with fingers open to the traditional measurement method (fingers closed). After adjusting for these additions, firefighters still have greater hand breadth than other groups (Table 9). Moreover, a comparison of finger dimensions between firefighters from this study and the best available Army finger anthropometry data (Greiner, 1991) showed significant differences for all finger dimensions for women and thumb dimensions and ring and pinky breadths for men. These results reaffirmed that general population statements on anthropometry may not be drawn from population-specific surveys and vice versa (Hsiao, Long, & Snyder, 2002) because anthropometric differences exist among specialized occupational groups, such as pianists (Wagner, 1988), truck drivers (Guan et al., 2012), and firefighters (Hsiao et al., 2014).

Structural firefighting glove fit is multidimensional, and hand/fingers size and breadth-to-length contrast are different in clusters by gender. Some manufacturers have included firefighter glove sizes designated for women in their product line. This current study demonstrated a need to include at least one or two glove sizes based on female hand models (i.e., Small and S-M sizes; Figure 4). Moreover, although literature has reported poor correlation between hand length and hand breadth (Gooderson, Knowles, & Gooderson, 1982; Vicinus, 1962), female firefighters who have a larger hand size tend to have a smaller hand/finger breadth-to-length ratio (Figure 3b). This study provides a step forward (i.e., including women hand model–based sizes and wide-palm sizes for men) for developing an improved glove sizing system for firefighters.

As presented in the Results section, there are discrepancies among existing firefighter glove size specifications, firefighter hand dimensions, and firefighter glove selection. The discrepancies may lead to several research questions. What are the acceptable tolerance ranges for palm and each finger for a single glove size? What is the most practical and accurate way to select the correct glove size? Although additional efforts are needed to address these questions, it is clear that firefighter hand dimensions and shape variations are greater than the corresponding ranges of the current sizing systems (Figures 2a, 2b, and 3b). The proposed seven-cluster system from this study contains Small, S-M, S-M Wide, Central/Medium, M-L Wide, M-L, and Large sizes. The Small size is more for female firefighters, whereas the S-M size can fit both female and male firefighters who have an elongated and small-to-median size hand. The remaining five sizes are in general for men with an understanding that some female firefighters will likely need the S-M Wide and Central (medium) sizes based on their hand size and shape (Figure 4). A comparison of the size ranges of each size between the proposed seven-size system (N = 919) and the sizes used by firefighters who reported that their gloves fit well (n = 434) shows that the proposed seven-size system has smaller ranges of standard deviation for almost all hand/finger dimensions for each size, except for a few dimensions in XS and XXXL sizes in which the sample sizes of three and five in the old/current system are too small for a meaningful comparison (Table 10). The proposed seven-cluster system offers a resolution for better accommodation and business value.

In this study we proposed a seven-cluster system as a step-forward resolution to improve the current firefighter glove sizing structure. This perspective assumed that each single glove size can accommodate a specific cluster of firefighters. There are at least three merits from this study. First, the method incorporated both hand/finger overall size and breadth-to-length contrast in the process (in lieu of a single dimension at a time), which is useful in lessening one of the primary problems with the current NFPA standard, that the accommodation ranges seem unrealistically large. Second, the number of sizes is less than or equal to the current practices, which does not create additional manufacturing and stocking burdens to the glove industry. Third, a representative hand model or a series of hand models from each cluster can be selected to make solid or digital hand prototypes for glove design and testing applications. The limitation of this study is that we do not know whether these new dimensional ranges for each cluster are the narrowest acceptable ranges. We are conducting studies on the acceptable tolerance ranges for palm and each finger for selected glove sizes. This effort will provide additional data for refining glove sizing schemes. A further cluster analysis using more or fewer clusters then can be performed iteratively to identify the best matched sizing plan. Until then, the proposed sizing modification represents the furthermost advancement in structural firefighting glove-sizing practices.