192
F. R. SENTI, M. J. COPLEP 9 S D
G . C . SUTTING
FIBROUS FROM GLOBULAR PROTEINS' F. R. S E X T I , A I . J. COPLEY,
ASL)
G. C S U T T I S G
Eastern Regioaal Research Laboratory,2 Philadelphia, Pennsylvania Received -1-ovember 22, 1944
Most proteins of current or potential industrial importance are in the molecular sense compact, almost globular structures as they occur in nature or as they are isolated by laboratory or commercial procedures. While the structure of none of the proteins is knoivn in detail, it is awidely held view that, fundamentally, the molecules are chainlike (1, 12, 2 5 ) but are normally maintained in a coiled configuration by interaction among the numerous polar groups. Opportunity for interaction between molecules is thus minimal. If the molecules were uncoiled and fixed in a linear form, the number of intermolecular bonds \vould be far greater, and the niechsnical properties of objects such as fibers, films, and plastics made from the altered proteins should be much improved. This is not merely conjectural, for filaments of the natural fibroiis proteins, silk and collagen, are far stronger and tougher than filaments of unoriented globular proteins. Casein fiber is now in commercial production, and experimentation is being conducted in several laboratories on the preparation of fiber from soybean and peanut proteins, zein, and other globular proteins. Fiber prepared in the conventional way from an nlkaline dispersion has in general been somewhat deficient in dry and wet strength and gives no evidence of molecular orientation in its diffraction pattern. It is the purpo.;e of this p;iper to describe experiments by which conversion of several globulnr proteins to the fibrous form has been accomplished and to give some of tlie pioperties of filaments made of the converted protein. Observations of Carothers and Hill (9), Sookne and Harris ( 2 7 ) , Mark (19), and others indicate that the length of an extended linear polymer molecule, ~ts measured the average degree of polymerization (D. P.) or molecular weight (XI. K.),rnuit exceed a critical rninimuni if fibers of usable pliability and strength are t o be prepared. For the polyester of w -hydroxydecanoic acid, Carothers and van S a t t a (10) found the limit to be bet;reen XI. 17. 10,000 and 15,000 or 11.P. about 60 to 90. 'l'his corresponds toalength of 800 trJ 1200 9.For cellulose and for polyamidcs the figures are roughly the sanie, niiile for hydrocarbons the D. 1'. for corresponding tensile strength is condcmbl:,- higher because of the wealme's of the internction betiwxi non-polar group.. In all these linear polymer moleculei, side groups are either completeiy absent or. if present, are small, are alikc, and u e spaced a t regular inteivals along the chain. On stretching, the molcctiles :ire mtdily oricnted and form yunsi-crystalline arrays in which segmenti of ndjncent molecules match one mother rather perfectly. Within 1 Par,.: of t h i i p.iper. n c i e piesciited at tlie Conference on Textile5 held a t Gibson Island, blaryland, July 12, 1914,and a t t h e 10Sth Meeting of tlie .inierican Chemical Society, S e n Tork City, SepteInbt.1 12, 1911. 2 One of the laboratories 01 the Bureau of .\giicultural arid Industrial Chemistry, Agricult u r a l ReseArch .\dmiiiistratioil I-nited States Dcp,irtnirtit of igricul'urc
FIBROUS FXO3I GLOBULdIZ PIIOTEIXS
193
these crystalline regions the potcnti:d energy tends to\\ arcl a minimum, imtl intermolecular cohesion is correspondingly increased. The molecular I\ eight of globular proteins ranges from about 15,000 to several million (29). -15 measured, it indicates the weight of the effective kinetic units in sedimentation, osmo , or diffusion, or the weight of crystallographically equivalent constituents in the iinit cell as determined h ~ s-iay . analysis. It does not necessarily gii e any indication of the molecular \\-eight, the degree of polymerization, and the maximum length of the polypeptide chain or chains contained in the molecule, and for our purposes tlicie are the significant quantitie?. For example, the molecular \\eight of hemoglobin is about 68,000. Of thiq, 2500 units are due to the four a\bociatcd haems. On treatnieiit even I\ ith a s mild a reagent as aqueous urea, hmioglobin splits into halves ot molecular \\eight approximately 34,000. From a partial structure analybis, Hoyes-Fatson and Perutz (8) h a m propowd that the hemoglobin molecule consiits of four substantially equal and parallel layers of polypeptide chains, 11ith tlie main chains folded in the plane of the layers arid the side chains extending a t right angles t o them. The position of the haem groups is nnknov n. It thiq model is correct, it i i evident that the upper and lower pairs of layers are not joined by a peptide or other strong bond. Whether the paired layer5 are themwlves linked by a strong covalent bond is not certain. Ii they are not, one uould expect to be able to split the pairs into halves ot rnolccular \\eight 17,000. ‘Thus far, there is no report that hemoglobin from the higher animals has becn dissociated into fragments of this size, but the hemoglobin of the cyclostomatn, containing one haem per molecule, has a normal neight of 1’7,000 (28). hlolecular-n eight measurements of globin derived iroiii normal (68,000 31. W.) henloglobin are not helpful, because globin ieadily forms loose association complexes and apparently it is the \\eight of the complex rather than the weight of the individual molecule that is usually found (24). Chibnall (11) has recently presented what seems to be good evidence that hemoglobin (68,000 M. JV.) contains sixteen peptide chains. These may or may not be of similar size and constitution. From what has been said abow, it is probable that not more than eight chains are firmly bonded together. If eight chains \\-ere joined by strong bonds adjacent to terminal amino groups t o make a continuous chain of \\-eight 34,000, the latter I\-ould have a D. P. of about 320, and if extended \ ~ o u l dbe 1050 8.long.3 This should make a satisfactorily strong, tough fiber. I t is important to note that in globular proteins about half the total \\eight ib in the side chains, so that a fully extended 34,000 31. JT.molecule o d d be much shorter than a molecule of rubber, 6-6 nylon, or polyvinyl alcohol of equal molecular weight, but nould be comp:trable in length to cellulose. The effect of smaller D. P. and chain length on interniolecular cohesion may, ho\rever, be partly or fully compensated for by the largc proportion of polar gioiips in both the main T h e average residue n e i g h t for globin i b 100 (Cliibnall (11)). Dividing 34,000 by 100 gives for the D P. a rou@ value of 320. The length of the extentled chain i b computed by iiiultiplying D P. hy 2 3 -4, the fiber icpcst distancc found in @-k(,r.itinand in glohular proteins converted t o t h o Fhrous Coiiii
and the G i J c chains. If tli:. ( ~ l i ~ i ip:ic,k i s i~ygilarly,:is t!icy d o in collagen, ~i strong ucture rei111 Should the eight polypcptick ciiaiiis iii tlie 1irmoglol)in hnlt'-molecule be lixiketl, not end-t o-end, h i t t!irongh c~rvss-1)ontl.;.tile length of fully estcrided chains \voulcl be corrcspoiidingly Aortcr, and conceivni)ly i i o usahly strong fiber could he mxde frorn 1ieniogloLiii reg:milc.+ of I i o \ ~perfectly the chains might be oriented a d packed. Taking an cstrciiic ('I~JC of sirtcrcn oqii:il polypeptide chains n-it,hout comlent. linkreating\:-ere done simultaneoiisly in a press. The heat treatment was usually given close to 100°C. Filaments of readily denntur:ible proteins such as lactoglobulin, serum albumin, the oil-seed globulins, and ovalbumin Irere simply kept in boiling water for 5 to 30 min. Filaments of casein, zein, and gliadin u w e conditioned in ivater r-apor at 0.5-1 .O atmosphere pressurc for several hours. During the heat treatment the filaments become tough and rubbery and so coherent that they may in general be stretched a considerable amount before fracture occurs. Changes in molecular structure that take place simultaneously \rill be discussed in the section on diffraction effects. Since the hc,zt-conditioned proteins arc decidedly viscous, application of shear stress provides the most obxrious means of producing molecular orientation. Stretching has been utilized almost exclusively, because it is a convenient technique :uid because it, is possible \vithout undue difficulty to initial observations has already been made (21). Of the several proteins t h u s far eraniined, ovalbumin has been outstanding in the variety of reagents found t h a t may substitute the thermal treatment. -4s judged by several tests, the niodc of interaction of protein n.it,h both heat a n d the cheinicals is similar. Ovalbumin filaments soaked in, for example, 'is per cent ethanol, or in boiling water become insoluble, elastic, nnd structurally bo composed that they can be stretched t o several times their iiiitial length. The s t r e t ~ l ~ efilaments d have much increased teiisile strength and are pliahlc w h ~ ndrx. 'I'hcy arc Imsitive-hirefriiigent and give $-keratin fiber patterns. A specimen of ovalbuinin surfncr c!r,naturcd iii a \\':iring Hleridor, wlieii strerclied, gave thep-keratin diffraction pattcrn. .in ynusual fcnturc of this pntTcI'I1 was the occurrence or: the equator of a reflection a t about 23 :I. Ovalbumin, disperseti in phenol arid s p u i ~i i i t o :i 1irrripit;Ltiiig !):if11 silrl! 2 5 ethnuo!, also gave u fiber p;ii fern after strctchiiig.
196
F. I t . SCSTI, .\I. J. C'OPLET .\SD G . C. STTTISG
prepare zpecinienb ruffkiciitl\- roliesi\ e t o permit btrotching. Stretching nas iisually (lone sloitly in steam oi in air approuiniately intirated 11 ith water vapor at xbout 100°C. I l r n n ratios (I). It.) or ratio> of final t o initial lengths of five or more \\-ere commonly obt:iinetl I n general, molecular orientation, as revealed by the diffraction pattern, increatied 11ith the tlran i.ltio under cornparable condition>. When spwimens iierc tlran 11 at room tempeiaturc or at an elevated temperature uith Y lo\\ pres~iircof \\-:iter vapor, :i w a l l chau ratio produced the same diffrrtc3tiorieffect\ a i :I muc.11 higher dran 1:ttio \\-hen the specimens were more highly pld~tirizecl. On titretching hi :til alkaline solution or in the vapor of a boiiing d u r a t c d solution of ci.eso1, clran- rJtio4 a.: high a. 25 have been obtained i{ithout any comparable increase in oiient_ttion. Attempt> to produce orientstion in the dwence of water have uniiormly failed. Specimen, prepared. for example, from intensively dried ovalbumin and cresol were stretched in cresol vapor at temperatures up to 150°C or more. The diffraction pattern of the stretched material conristed of t\vo full, diffuse rings. Since the orientation proc appear3 to be rather generally applicable to globular proteins, it is perhaps not surprising that proteins of various degrees of purity and from varioiis miirce- should when oiknted give comparable diffraction effect>.: In our euperience there iq. Iioivevcr, x marked variation in the ease with T\ hich the tliffei ent proteins tirid protein preparations may be crystallized and oriented. (The \\or11 ciystnllizetl i.j used in the restricted sense that the s-ray reflection9 sharpen :ind become more niimei ou-, indicating increased regularity or order in the spsti:il arr:ingement of the scattering matter.) Generally speaking, the protein3 ordinarily clawed a > x. The polypeptide c1):iin.s are parallel t o J : tlie 10 -1.spacing is approsinxitely along z ; the 5.G -1.spacing, approsini::tt.Iy d o n g y; :ml the 3.7 -1.spacing, a diagonal spacing perpendicular to the direction 2. In unstretched filaments t,lie platelets have random orientation A\.
201
FIBROCS FROM GLOBULAR PROTEISS
(figure 32). Upon application of stress the first tendency will be for the plane
zy to orient in the direction of stress, that is, the fiber axis, but with no preferential orientation of x or y of the various platelets (figure 3b). Orientation of the planes xy in the direction of stress requires the lattice planes with the 10 A. spacing to be parallel to the fiber asis and hence to reflect on the equator. Since the planes q have random orientations arotnd a perpendicular to the fiber asis, the lattice plane; of spacing -1.6 and 3 7 -1.11ill reflect n-ith uniforni intensity through 360" on the x-ray photograph. Upon further stretching (figure 3 c ) , the longest axis z of the platelets TI ill turn ton 3rd the direction of .tress, orienting the 4.6 A. planes parallel to the fiber axi.;, and callsing the corresponding reflection on the x-ray @otograph to lie arced on the equator. Similarly, the diagonal planes of 3.7 -1.cpacing I\ ill give four intensity maxima at approsiniately 4.5". -iccordin~to tlie structural scheme proposed b>-&%stburyand Woods ( 5 ) , the 10 and 4.G spacing. are lateral qpncings of the polypeptide chains and are a[>A\.
b
a
C
FIG.3. Orientation of crystallites during stretching: (a) initial randoin orientation; ( b ) uniplanar orientation; (e) uniaxial orientation.
prosimately a t right angles to each other. Froni changes observed on swelling, Astbury and Lomas (3) concluded that the 10 .: spacings are in the direction of the side chains, while the -l.G .i!- spacing represents the ieparatioii of the chains in the plane of the zigzag of the main chain, as illustrated in figure 4. From the orientation sequence, then, the crystallite< must be much longer in the direction of the backbone spacing than in the direction of the side-chain spacing. This conclusion is supported by the breadth of the corresponding reflections, the 10 reflection being much brondrr thnn the 4.G -1.reflection. Assuming equal lattice perfection in the tn-o direction\. this indicates that the tallite iq much smaller in the 10 9. or x direction. Aldditionalevidence is supplied by ,specimens TI Iiicli hRre been rolled to give biaxial orientation of the crystallites. X-ray pattern. of siich spcciinenq takcn I\ ith the beam painllel and a t right angle5 to tlie plane of rolling are -ho\\-nj11 figure 5 . It 1' seen that thc crystallites orient ,SO that the plnneb nith the 10 5 spacing art' parallel t o the plane of i d l i n g , that is, the z direction of the crystalhtr- ic pcr-
x.
202
F. R. SEXTI, &I. J. COPLEY A N D G. C. NUTTISG
a
M
I
I
-0-2-Q) -0--Z-cg,-0-Z-@-0-
aI
I
I
I
I
I
a a
a
I
I
I
4
I
I
r
aI
I
r
-0-2-e3 -0-2-(9-0 -2-@-0I I QL
a
I'Y
a
I-0-z-e-o-r
I I
I I
-2--(ID-0-2-@
I
I
I
I
a
of
o = a z\ I a I 00\ , * , 0cl)\ 0 0, *,*, z 0 e> P I I I a / \ I o aa= I 8 I
8
I I
\
\ /
6 I
/
a: I
=f
203
FIBROUS FROJI GLOBULAR PROTEISS
pendicular to the plane of rolling, while the z and u directions lie in this plane. This, again, is consistent jrith the relative lengths z > y > > z . The crystallites thus have a pronounced tendency to grow in the direction in TI hicli the polypeptide chains are bonded laterally by hydrogen bridges between
‘Y-H
/-
\
/c=o
nl’d
groups of adjacent chains. The regular occurrence of these grotips
along the chain gives opportunity for regular packing of chains in the direction of the hydrogen bridges. On the other hand, the packing in the direction of the side chain< TI ill be influenced by the variable length and nature of the qide chain n
I
f
w
\
I
b
3,
C
FIG.5 . Diffraction patterns of biaxially oriented ovalhumin.
(a) X-ray beam perpendicular t o the fiber axis, parallel to the plane of rolling: ib) besni perpendicular t o both fiber axis and plane of rolling; ( e ) beam parallel t o fiber :+\is. Plane of rolling is vertical.
f
\
L 1
\ W
a b FIG.6. The tivo types of fiber pattern exhibited by oriented globular proteins.
(31
the
“@-keratin” patterns; (b) the “egg-n-hite” pattern.
from the different amino acid residues, and consequently the order n-ill be much less perfect and will extend a shorter distance. The 3.3 9. “fiber repeat” reflection is normally vieak. It becomes proiniiient only in rather well-oriented specimens, even though the repeating units lie in the direction of the longest diniension of the crystallites. The tendency of the crystallites to grov in the y direction may explain another type of diffraction pattern frequently observed in clran-n protein filaments. This pattern, illustrated in figure G , 11 as first obqer\-ecl by Xstbury, Dickinson, and Bailey (2) in stretched films of poached egg 71 hite. The distinguishing feature of this piitteiri i. that the 4.G -4. reflection i. arced on the meridian rather than on the equator. the interplnnar Fpacingr oi the three principal reflections being the same RS in the 3-keratii? type. -litbury, Jlickinwn, and 13Llilc>>-( 2 ’
204
F. R . S E S T I , If. J . COPLEY .IND G. C. SUTTING
report a faint reflection of 3.3 *I.spacing on the equator, corresponding to the repeat distance along the polypeptide chain, but we hal-e never observed this although we have esaminecl dozens of photographs showing this egg-white pattern. The simplest esplanation of this photograph is that the crystallites have developed such that 'y > n: o> x . Hence, upon stretching, y will orient along tke fiber axis, causing the 4.G A. reflections to fall on the meridian, while the 10 A. reflection occurs on the equator as before. This pattern has been obtained from ovalbumin, casein, zein, peanut protein, hemoglobin, arid edestin. I t usually occurs if a fiber is given relatively low extension after a relatively short period of heat-conditioning. For example, it appears when ovalbumin filaments are immersed in boiling water for 2 to 5 min. and are then stretched 50 to 85 per cent. At greater elongations this pattern is transformed to the 0-keratin pattern. The reverse transformation of the @-keratin pattern into the egg-n-hite pattern has never been observed. Although many oriented protein filaments h a w been contracted in hot rrater or steam to give disoriented /3-keratin, an intermediate form giving the egg-white pattern has not occurred. BIREFRINGENCE
X-ray diffraction provides the best method of gaining information about the orientation and packing of the molecules in the crystalline regions of a fiber, but i t gives little direct information concerning the alignment of chains in the less-ordered regions. Optical and swelling anisotropy do, hoivever, provide a rough measure of orientation of intercrystallite chains. The refractive index of a substance for plane polarized light depends on the polarizability of the atomic groups in the direction of the plane of the electric vector. When a high polymeric substance is stretched, there is an alignment of molecular groups which usually results in greater polarizability along the direction of stretch than a t right angle. to it, with a consequent difference in refractive index in these two directions. Birefringence has the advantage that it can detect orientation in substances which do not produce any easily observable x-ray diffraction effect and are thus seemingly amorphous. On the other hand, calculations by Treloar (30) indicate that the birefringence is insensitive to small departures from perfect orientation of crystallites along a given direction. X comparative study of birefringence and diffraction effects has been made on ovalbumin fibers heat-denatured and stretched in steam. Birefringence was computed from the retardation and the diameter of the fibers. The retardation was measured by means of an uncalibrated quartz xedge. Birefringence changes n-ith draw ratio, as shon-n in figure 7 . Xt I> R. = 2, diffraction patterns of the same fibers sh;w marked arcing of the 10 Ai.reflection, but only T-ery slight arcing of the 4.6 .$. reflection. This corresponds t o approximately uniplanar orientation of crystallites. as augge5ted above, and judging from the x-ray pattern alone the filaments should he optically nearly isotropic. The birefringence is G X loL3,however, suggesting that there is preferred orientation of polypeptide chains in the direction of itwtch. Treloar (30) predicts that in this region of low
305
FIBROUS FROM GLOBUL.111 PBOTEISS
orientation the birefringence n-ill change most rapidly. Part of the birefringence may be contributed by preferentially oriented intercrystallite chains that do not appreciably affect the diffraction pattern. This wou!d be analogous to the behavior of polymerized methyl methylacrylate when stretched. The polymer molecules are evidently linear, but it is difficult or impossible to pack them into regular arrays in space. -4lthough when stretched the material becomes birefringent, the diffraction pattern shows very little asymmetry (15, 23). Similarly, a number of samples of casein textile fiber prepared in this laboratory by precipitation from an alkaline dispersion, followed by stretching and formaldehyde hardening, had birefringence ranging from 1 to 10 X low3n-ithout any indicntion of arcing on the diffraction patterns.
I
2
3 DRAW
4
5 RAT 10
6
7
FIG.7 . Change of birefringence of ov:ilbuniiii lvitli drnn. w t i o
With increasing stretch, the birefringence of oralbulin increase+, idiile the arcs on the x-ray pattern decrease in angular width and increase in intensity. -it D. R. = ithe birefricgence is 14 x 10-3, and the slope of the curve of figure T indicates that it has not yet attained its rnasimum value. It niight be supposed t'liat protein preparations giving the egg-n-hite type of fiber pattern would be negative-birefringent. The refractive index i.5 greatest along the direction of the extended peptide chains, and t1ii.Qdirection in the crystallites iyom which the fiber pat,tern arises is perpendicular t,o the fiber axis. -ICtually the birefringence of all the several specimens examined was positive, nnd of the order oi The folloT.r-ing may be offered in explanation of this anomaly: By simple visual comparison of the intensity of arcs and rings, it is clear that the proportion of crystalline matter is not high. Furthermore, the diffraction pattern obtained with the beam parallel to the fiber asis has full rings of comparable intensit,y at, 4.6 and 10 &8.. In the diffraction patt.ern of the crystallites alone, the 4.6 -61. 1.in.g would IN missing. Conseqiiently, the negative birefringcnrc can-
200
F. R . SESTI, 11. J. COPLEP AND G. C . NUTTISG 1
tributed to the total birefringence by the crystallites is small and is apparently overweighed by the poqitive contribution of the non-crystalline chains oriented parallel to the fiber axis. High optical anisotropy TI ithout corresponding x-ray effects has a130 been found in fibers drawn irom surface films. At a water-air interface, soluble globular proteinq in general unfold t o form a monolayer. This, on compression between parallel barrierb, iq anisotropic, and from i t filaments may usually be drawn (17). ,\ccepting the vien of Langmuir (16) that protein monolayers are composed principally oi long polypeptide chains, it might be evpected that filament"?pulled from such monolayers n-ould contain the chains in a linearly extended parallel array. IT-e have prepared fibers suitable for examination from monolayer$ of casein and ovalbumin spread on a 0.003 -If citric acid-phosphate buffer at pH -1.7. At a preqqure of approximately 25 dynes per centimeter, fibers were formed by sloi\ ly n itlidran ing a 0.5-mni. platinum n ire from the film-covered surface When maintained taut until air dry, many of the fibers had high 1)o"itive birefringence, a mn\;imum value of 7 X being recorded. -4h n d l e of fine ovalbumin fiber$ ha\ing a rather uniform birefringence of about 2 X lo-? n a< examined in R niicro-diffraction camera. The diffraction pattern consisted of full ling&,n i t h no +uggestion of arcing. The combined evidence of birefringence and x-ray diffraction I\ o d d qeein to indicate that in the fibers there is a high degree of preferred orientation of chain molecules which are imperfectly packed. Because of the extreme smallness of the fibers, no attempt was made either t o improTe the .uati:il order by an annealing process or to measure the tensile strength. s\VELLISG MEASCREJLENTS
; I Ytretched protein hber dried under tension generally contracts in length nheii immersed in witer. and contracts further upon drying. If the fiber iy again immersed in ivater, both length and diameter increase. The contraction in length and the increase in diameter upon the first immersion in water diminish as the 3tretch given the fiber increases. For example a t D. R. = 2 a n unhardened ovalbumin fiber contracted 14 per cent in length and increased 33 per cent in diameter, \\ hile at D. 1%. = 8 the corresponding values were 3 per cent and 19 per cent.e Similarly, the reversible dimensional changes in a completely relaxed fiber decrease uith elongation. I t D. R. = 2 a relaxed fiber on wetting swelled 12 per cent in length and 20 per cent in diameter, while a t D. R. = 8 the value? 11 ere 4 ancl 13 per cent. The fiber thus appear^ to he stabilized ton ard water by stretching. If atretching afiected only the orientation of evisting crystallites, we should evpect that lateral elling oulil increase ancl longitudinal sn-elling decrease. Tnstead, both longitudinal anti lateral a n elling decrease. This suggests that the niolecular packing iq imprm ed by .tretching, that iq, crystalline areas are produced. Swelling anisotropj- i+ defined as the iatio of the increasc in diameter to the mcreace in length ot n Eber on inimerbion in a, swelling liquid. For the ovalbumin I
( U
:lie i \ e r fibs, 011 drying, further ilimcnsioiisl changes occur
20;
FIBROUS FROM GLORL7L.U1 PROTEISS
fiber- of D. R. = 8 mentioned above, its value iq slight!!- greater than 3 . For cotton, wool, and silk, the value is about 10, whereas for regenerated celllilo-e it i.: about 6 t o 8 (31). TESSILC S P R E S G T H
expected, tensile strength is increased upon conr-er;ion of globula.:. protein? t o the oriented fibrous form. Tensile nieasurementq have been made tor the
most part on coarse filaments, approximately 150-300 microns in diameter Single-filament breaking loads were measured on a Scott Type I-P-2 inclinetlpiane testing machine. Specimens were kept in the testing room at 70°F. and 65 per cent relative humidity a t least 16 hr. before breaking, Initial diameters were used in the computations. Strengths reported are in general the average strength of ten or more fibers. Air-dried extruded filaments of all the proteins nere so weak and brittle that tensile meawrements were impracticable, but after the heat treatment the -trengthg 11 as 4.5 t o 6.7 kg. per nim.2 Upon orientation, strengths were raised to the following values: casein, 18.6 kg. per mni.?, pumpkin-seed globulin, 16.2; hemoglobin, 11.6; lactoglobulin, 12.8; zein, 17.6; oivalbumin, 20.9. Individual filament strengths as high as 30 and 34 kg. per mm.? have been measured on )eelmen.: of casein and ovalbumin, respectively. ;111 the oriented preparations gave 3-keratin-like fiber patterns. It seems clear, hon e1 el, that by 110 means all the protein even in our best specimens is in the form of oriented cryytallites. although \ \ e knon of no experiment by which the proportion; of crystalline and noncrystalline material may be precisely judged. On inimeryion in water, dimensional changes and change.; in biretringence suggest that part of the protein in stretched filaments i- unfolded and oriented but i z not crystallized, and thus seems amorphous to x-ray-. We are unable to e3tiruAite the propor tion of molecules n hich hare substantially the \mie configurntion n- in the original protein. If the protein chain is 1000 d.or so long \\-hen fullj- c\tended. t 1 0 01 inole portion. of the chain could pack in orderly arrays \\ itli m,itcliing iegment; of other protein molecules to form crystallites of a not improbable -ize-bt u- -xy 100200 !, Intercrj-stallite regions, although composed from the \mie chain5 iorniing the crystallites, may be incapable of forming a regular netnork in -pice Thi- may be caused by chain flexibility, 1-ariable chain length betn een cry2t:iIlite:\:en Torli Academy of Sciences in S e w York City, January S and 9, 1943.
