240 JOURNAL OF COSMETIC SCIENCE with the thumb while pulling the comb through the hair, placing the major stress against the load cell and the back and side of the comb. These data (Table II) show that thumb pressure is an important variable in the actual compression loads generated during combing. Furthermore, it is best (providing lowest compression loads) to use the thumb to ensure that all hairs remain in the comb teeth rather than to apply pressure that provides higher compression loads and more damage to the hair. These data also show that the size of the tress or the amount of hair in the comb also contributes to compression loads during combing. Very high compression loads were achieved in these experiments. These compression loads are due primarily to the thickness of the tress, the comb teeth spacing, which in this case is large, and how hard one presses the hair against the comb or the compression cell. These variables are all difficult to control among different qualitative combers and undoubtedly account for variations seen in qualitative combing experimentation. These factors determine the number of fibers that fit between the comb teeth and also how the tress is held against the comb during combing, and these loads are distributed over a large number of fibers and not in a uniform manner, suggesting that for some hair-on-hair contacts, very high compression loads are encountered. ESTIMATING THE COMPRESSION LOAD AGAINST SINGLE HAIRS DURING COMBING To estimate compression loads on single hairs during combing, hair-on-comb compres­ sion loads were measured in a snag underneath the advancing comb held in a vise to control the hair fibers in contact with a specific part (of known size) of the compression cell. A miniature trees of very curly hair (Caucasian hair steam-set to simulate African hair) was used, and only a small number of hairs, approximately 100, were inserted on each side of the comb tooth containing the compression cell to provide a snag and control the hair against the load cell button only. In this experiment the maximum total compression load for 17 determinations varied from 271 to 1867 grams. With two primary assumptions, the average maximum compression load between the comb and each single hair was estimated at 39.5 grams per fiber, and it varied from 16 to 110 grams per fiber. The assumptions are: (a) the hairs are perfectly aligned against the button of the load cell (1524 microns wide) and (6) each hair is assumed to take up 90 microns of space (70-micron average hair diameter plus 10 microns on each side for hair crimping and imperfect packing). Therefore, 17 hairs are assumed to be in direct contact with the load cell. During these runs, for data to be recorded, a uniform layer of hairs against only the button of the load cell had to be apparent, and hairs could not be in contact with the rest of the Table II Average Combing and Compression Loads for Undamaged Wavy Caucasian Hair Tress size Thumb loosely Against back of comb 2-gram 22 ± 11 80 ± 50 4-gram 56 ± 20 290 ± 82 6-gram 205 ± 82 889 ± 316 Date is expressed in gram load within the 95% confidence limit. Against cell 449 ± 111 602 ± 196 994±231

PATHWAYS OF HAIR BREAKAGE 241 load cell because the entire compression cell surface provides compression readings and the button could be used to gauge the actual amount of contact between the compression cell and the hair. It is also assumed that the compression loads are distributed uniformly over these hairs. These experiments demonstrate that compression loads between hair fibers and comb surfaces during combing can be very high. One can assume that the average compression load for a hair-on-hair interaction in a difficult snag would be of the same order per fiber or in some cases even higher, leading to very high compression forces per unit area in actual contact for one hair fiber against another. Therefore, for hair breakage during combing or brushing, compression forces are in­ volved in addition to extension forces, and this fact is consistent with all three of the previously suggested pathways for breakage. Furthermore, these compression forces can be very high, and with rubbing can lead to extensive cuticle disruption (3 ). Yet com­ pression forces are not considered in ordinary tensile testing of human hair fibers. IMPACT LOADING EXPERIMENT Most tensile testing conditions employ very slow strain rates, of the order of 0.25 cm per minute. Under these conditions, hairs generally stretch about 40% to 60% of their length before breaking. However, during combing, hairs do not stretch to such lengths before breaking therefore, another important variable is likely to be strain rate differ­ ences in combing versus those in tensile testing. During combing, strain rates can be very rapid, at least an order of magnitude faster than in normal tensile testing, and combing more closely simulates impact loading than slow stretching. Therefore, to more closely simulate the action of combing, experiments on hair fiber breakage using impact loading conditions were examined. BREAKAGE OF HAIR OVER HAIR VS HAIR OVER COMB TOOTH Hair loop over comb tooth vs hair loop over hair loop. Hair fiber loops were made as described in the Experimental section and the weighted loop (51-gm load) was impact loaded over a large thick-comb tooth (varying from 1800 to 1400 microns thickness). Hair breakage generally did not occur on the first impact (see Table III). However, when the comb tooth was thinner (varying from 1245 to 940 microns thickness), using one of the fine teeth on this same comb, breakage occurred with fewer impacts using the same 51-gm load (Table IV). Yet, when one hair fiber loop was impacted over another hair fiber loop Table III Impact Loading a Hair Fiber Loop Over a Thick-Comb Tooth (-1800 µ) Using a 51-Gram Load 4 Fibers broke on impact 1, near contact site 2 Fibers broke on impact 2, at contact site 2 Fibers broke on impact 3, not at contact site 3 Fibers broke on impact 4, at contact site 1 Fiber did not break after 10 impacts (36 Total impacts and 11 broken hairs)