46 ,JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS Table V Fracture Pattern Distribution of Failures in Fatigue Test to 11 kc Under Various Loads (Nonsurvivors) (Fracture at 65% RH) Fracture type (%) Load Number of (g) Specimens Smooth Step Angle Split Fibrillated Untreated 0 36 6 56 6 10 22 10 17 0 29 0 18 53 20 16 19 0 6 0 75 30 20 17 7 30 13 33 40 45 2 36 29 11 22 SR 0 50 40 32 4 8 16 10 19 53 11 5 5 26 20 11 36 9 27 9 18 30 34 21 38 29 6 6 40 62 24 50 10 8 8 TGA 0 50 40 34 6 10 10 10 11 27 9 0 9 55 20 15 27 13 13 7 40 30 30 17 7 30 13 30 40 58 43 26 5 7 19 differences in the angle, split, and fibrillated ends as a result of treatment are not significant. The increase in the number of smooth fractures seems to be due to enhanced relaxation at the crack tip, which prevents stress concentrations that can be directed along the axis of the fiber as discussed earlier (4). The tendency towards enhanced relaxation is reflected in increased extension at break for superrelaxed and thioglycolate-treated fibers lower modulus and breaking stress is also observed for these fibers. Fatigued fibers show a lower percentage of premature failures than unfatigued fibers, suggesting that the fatigue test eliminates a considerable number of damaged fibers which would have failed at low extension if they had survived the fatigue test. The fact that unfatigued fibers show a slightly greater tendency to fibrillate than fatigued fibers is also probably due to the elimination of damaged fibers in the fatigue test this interpretation is supported by the decrease in the number of fibrillated ends with increasing load, which would appear in the fracture pattern distribution of nonsurvivors (Table V). Fracture pattern distributions of fibers that failed in the fatigue tests are shown in Table V. Fiber treatment and fatiguing seem to affect the number of smooth, angle, and fibrillated fracture ends more than step and split ends, which do not show any particular trend. It seems that fibrillated ends result from a combination of preexisting damage to the cuticle and fatigue (tensile and torsional). The fact that fatigue test survivors do not fracture with fibrillated ends suggests that fatiguing to this extent does not damage or loosen the internal structure of the fiber to cause fibrillated ends.

EFFECTS OF TREATMENTS ON NEGROID HAIR 47 Tension Resulton• • Torsion Reo•iøn -torsional fatigue Torsion I• I Resultant Tension Figure 3. Schematic diagram of the tensile fatiguing of a twisted fiber. In Table V the fraction of fibrillated ends goes through a maximum with fatiguing load, and the number of angle fractures seems to increase with increase in fatiguing load. This is probably due to the torsional component in the tensile fatigue experiments. This effect in the twisted region of the fiber is shown schematically in Figure 3. It can be seen that extending the fiber reduces the angle of twist and hence the magnitude of the torsional component. Since loads of 30 and 40 g are close to the yield value (Figure 4), fibers extend significantly to reduce the angle of twist, resulting in a lower torsional component hence the severity of torsional fatigue is reduced considerably at high load levels. This seems to result in a lower fraction of fibrillated ends and a higher fraction of angle fractures. To understand this aspect better, it would be useful to study fracture patterns as a function of cycles to failure on a much larger sample. This might show whether fibrillated ends occur specifically at high cycles to failure, in which case they could be attributed to fatigue effects. The fact that fatigued fibers show higher fractions of fibrillated ends than unfatigued fibers shows clearly that fatigue plays a role in the fibrillation mechanism. However, fatigue seems to be effective only in combination with some preexisting damage to the fiber structure. From the data of Table V it is clear that fibers superrelaxed with alkali resist fibrillation more than untreated and thioglycolate-treated fibers. The chemical effects of treatments of Caucasian hair with alkali and thioglycolate have been studied by Chao eta/. (5). They found that alkali-treated hair has much higher levels of lysinoalanine and lan- thionine than thioglycolate-treated hair, thus producing a considerably higher number of stable crosslinks. These authors also studied the number of SH groups by incorpo- rating 3H-iodoacetic acid into the cuticle and cortex of alkali-relaxed and thioglycolic acid-treated fiber. The concentration of SH groups was found to be 10 times higher in