OREGON FOREST PRODUCTS L.AaDEATC! 9Y LIBRARY THE EFFECT OF CHANGES IN TH E EQIJILI13RIIJM RELATIVE VAPO R PRESSURE UPON THE CAPILLAR Y STRUCTURE OF WOO D October 193 5 No. 81075 UNITED STATES DEPARTMENT OF AGRICULTUR E FOREST SERVIC E FOREST PRODUCTS LABORATOR Y Madison, Wisconsi n In Cooperation with the University of Wisconsin F,OU LFHE T .nqn U a+JRATOR g EFFECT OF CHANGES Iiv TET EQUILIBRIUM RELATIVE VAPO R PRESS i `, ATTHE CAPILLARY STRUCTURE OF WOOD , By ALFRED J . STAiM, Chemis t Introductio n E The rate of penetration of various liquids into :.soft g oods ha s and by Sutherland, Johnston, an d been studied by Johnston and Maass Maas:, (12) from the standpoint of pulping . They-found the rate o f penetration of transverse heartwood sections with thicknesses greater tha n the fiber length to be about 100 times greater than the 'penetration of lik e thicknesses in the radial and tangential directions . Unseasoned sapwoo d was as much as 200 times more permeable than the corresponding unseasone d heartwood.. An increase in pressure caused . a slightly greater rate o f penetration at higher pressures than would be expected if it were directl y proportional to the pressure . This the authors believe to be due to a stretching of the pit membranes at higher pressures, resulting in an increase in the size of the openings . The use of back pressures had n o apparent effect . The rate of penetration in all cases decreased wit h time to a final equilibrium rate . Presoaking di d, not hasten the attain ment of an equilibrium rate . (0 The capillary structure of wood has been recently studied b y Buclaaan, Schmitz, and Gortner (2) . They found the rate of penetration of transverse sections of softwoods to water, aqueous solutions, an d various organic liquids to decrease with time to a final equilibrium value . The change in the rate of penetration of water caused by the addition o f electrolytes was not a function of the viscosity . This they believe t o be due to electrokinetic ; effects which modify the resistance to flow . They also found the rate of penetration of benzene to decrease with increasin g moisture content of the wood, indicating that the swelling of wood cause s a decrease, in the size of the effective openings . 1 `Presented before the Colloid. Division, American Chemical Society, at It s 89th meeting in New ?or City, April 22-26, 1 935 . R1075 The author has determined the effective size of the openings i n wood through which flow occurs by several different physical methods : The effective capillar=y cross section of various softwood sections wa s determined by means of electro-osmotic flow (7 ) 8) and by measuring th e electrical conductivity of salt solutions filling the capillary structur e ( ) . These data, combined with .data obtained from pressure-permeabilit y . measurements to water, made possible the calculation of the average effective capillary radius .( 7 ) , ) . Measurements of the air pressur e ,-required to overcome the effect of the surface tension of water held i n the caillary structure furnished data for the calculation of the maximu m effective capillary radius (7; 8, .2) . The average effective capillar y radius was found to vary from 10 to 70 mid for the heartwood of differen t species tested and from 180 to 2,000 ii for the sapwood . The heartwoo d values are of the same order of magnitude as the size of the opening s found in synthetic membranes by other investigators 6 , )4) . The maximum effective capillary radius was found to vary from CFO to 120 n7a for th e heartwood of several s pecies and from 180 to .11, 000 mid for the sapwood . The maximum effective openings are thus about two to five times th e average in size . The data indicate, that these effective openings ar e either normal pores in the pit membranes or checks across the membranes . In the case of thin transverse sections less than the minimum fiber lengt h in thickness, the flow is through the open fiber cavities . . Measurement s made on these sections gave capillary sizes of .the same order of size as . the fiber cavity dimensions determined microscopically . The only available data to give an indication of the effect o f shrinking and swelling of wood upon the effective capillary sizes are thos e of Buckman ; Schmitz, and 'Gartner (2) which ' indicate that the effective , . capillaries increase in size upon drying of the wood . More complete information along this -line should be of considerable value in . determinin g and understanding seasoning, preservation, and pul3jing practice . E:erimental Procedure. An attempt was first made to determine-the equilibriu m -permeability of wood sections to air of different relative vapor presmtre s using the differential pressure drop apparatus previously described (z~) • It was found, however, that when green sections were securely clampe d the apparatus so as to avoid leakage, the 'sections subsequently choc .:ea . when brought to equilibrium with air of lol-T relative vapor pressures . Th e apparatus was hence modified as shown in figure 1 . It consisted esse n tially of three compartments, 1 and 2 separated by the wood section and 2 and 3 separated by the standard capillary tube C the same as in the previou s apparatus . Manometers were provided to determine the prase are drop betwee n compartments 1 and 2,and 2 and 3 . The means of holding the sections wa s x1075 - I .1 hl 0•1r F Figure 1 .--Low-pressure permeability a pparatus usect for making, measurements on transverse sections of )- .root? . modified as follows : Holes were drilled in one end of cylindrical piece s of wood in the fiber direction with a centerless bit, the bottom servin g as a transverse section . A soft rubber stopper R connected the section t o the apparatus . It shored no tendency to lobsen or exert an appreciabl e stress on the section over a complete moisture change cycle . The woo d section was enclosed in a large cylindrical glass tube with rubber stopper s at each end . A stopcock S was provided in compartment 2 to prevent th e passage of air through the wood section during .the course of bringing i t to moisture equilibrium . During this process the air entering at A, ci iculated around the section, and was then discharged at B throug h stopcock S' . This was done to cause humidification to occur from th e outside of the specimen so as to more nearly duplicate normal drying an d reabsorption conditions . An electric heating coil was wrapped about th e exit tube B to prevent condensation of water from air at a high relative . vapor pressure when being discharged from the apparatus . The apparatus wa s held at a temperature'of )40° C . in a . thermostatically controlled water bath . The air waa humidified by bubbling through saturated salt solutions i n 10-inch high glass jars filled with glass beads . About an inch of mercur y was placed in the bottom of the jars and the air first bubbled through thi s to prevent salt from working back into the jet . The air passed through a trap filled with glass wool to remove any traces of entrained spray befor e entering the apparatus . The relative vapor pressures given -in table 1 wer e obtained from gravimetric determinations of the rmoi .sture content of the ai r at approximately the same rate of flow as that used in the experiments . r This apparatus proved satisfactory for determining th e permeability of transverse sections . It could not be used, however, fo r radial and tangential flow as practicall y, all the air would pass . through the end-grain part of the cylindrical wall of the wood section rather tha n through the end, In the case of the transverse sections the passage of ai r through other than the end-grain ends was entirely negligible (1 percent or . less) . Further, this glass apparatus could not be used for other than th e transverse sections because it would not stand the pressures required fo r obtaining measurable permeabilities with some of the resistant sections . It was thus necessary to devise another means of making the measurement s using a more rugged apparatus . It was found that sections clamped . in the old type of flus h rubber--gasketted flange,' when carefully marked, could be removed from th e clamp and reinserted in practically the same position with a variation i n the pressure drop reading s r of not more than 2 to 3 percent . It wras henc e decided to use this type of flange clamp and remove the sections and pre humidify them to approximately the equilibrium relative vapor pressure o f the air to be used in the next permeability test . In this way checkin g .of the sections was entirely avoided . Although the source of error i n these measurements was somewhat greater than in those made with th e •apparatus of figure 1, there was the distinct advantage of being able to . make measurements on one series of sections while another series wa s being prehumidified . R 1 075 Table 1 .--Relative vapor pressuresobtained with different saturate d salt solutions Temperature Material :•Relative vapor pressur e -------------------------- --------- -------- ----------------------- 0 ti C. Distilled water : 40 96 . 3 Saturated BaC12 : 40 SG . 7 Saturated NaCl : CIO 73 .1 Saturated gnC12 40 50 . 5 Saturated MgC12 . . . 4o 3 1`0 0 4o 11 . 7 Saturated LiCl . •: 0.0 Distilled water : 23-2 5 9g . 0 Saturated BaC12 : 23-25 g 9! 3 Saturated NaCl : 23-25 74 . 5 Saturated MgCl2 : 23-2 5 31 . 0 P205 23 -2 5 0 .0 R1075 - 4- e4 • More substantial humidifiers were also •necessaxy for these , measurements . They were .mado of 1S-inch lengths of 1-inch galvanized iro n pipe . Pritted glass discs were clamped to the lower end of the pipe wit h a 1 by 1/4 inch reducer with rubber gaskets•between . Pieces of 1/4-inc h . galvanized pipe were threaded to the reducer .and bent in a U shape . Similar reducer s' with short 1/4-inch nipples were screwed into the uppe r ends of the 1-inch •apes . The humidifiers were coated inside with a mixture of half beeswax and half rosin wax to decrease corrosion . They were then half filled with water or . saturated salt solution s'. Air pressure applied to the bent intake pipe was broken up into fine bubbles in passin g through the fritted glass disc, and thence through the saturated sal t solution. . To prevent the salt from clogging the 'ritted glass discs, ai r was always passed through a water prehumidifier first . The salt 'solution s thus always reduced the relative vapor pressure of the air rather that . raised it'. A similar humidifier with an inside 1/4-inch standpipe served ' as a trap . Glass' wool was inserted in the upper part of it to remove an y traces of (pray . Five humidifier tubes and one trap were assembled-on a rack . They were used only at room temperature, 23°•to 25° C . The relative vapor pressures obtained with thq e humidifiers are give n . in table • 1 . Th e sections were prehumidified in humidity rooms held at SO° F . and 30, 75 , 90, and 97 percent relative humidity, respectively, and in a desiccato r over P 0 for at least a week . The equilibrium relative vapor pressur e change 2 In the apparatus was so small that readings could be made a s quickly as pressure equilibrium was established . The velocity of the flow of air through .both the standar d capillary tube and the wood sections can be expressed by Peiseuill e t s equation : i 4 V _ ~' rcPc . ° E q 1c (for standard capillary tube c) 4 rw pvi V1v = HQ,"A(for the wood section ) S lt , (1 ) (2) ' in which Vc and•Vw are the velocities of flow through the capillary an d through t wood section, rc and rw are the radius of the capillary and th e ' average effective radius (to be more exact, the fourth root of the averag e fourth power) of the` wood capillaries ., lc and ltiq are the correspondin g lengths and Pc and Pr the corresponding pressure drops, 1 the viscosit y of the ai r? Iti the number of .effective caaillarien in the we0a sections i n parallel per unit of cross section and 2, the effective cross section, al l expressed in centimeter-gram-second units . When the standard capillary and the wood section are connected in series as in this investigatio n x1075 f. (3 ) vc and iARv = 1 r Pc iw lcc Pw b (4 ) Both N and rtiw are unmiown so that in order to determine either, farthe r data are necessary (8, .) . During the course of the shrinking and swellin g of wood, N will, however, remain constant unless checking occurs . The pressure drop ratio Pc and the standard pressure drop rati o Pw 1. should thus be proportional to the fourth power of the capillary radii . The change in size of the capillary openings in wood have hence bee n expressed in terns of the pressure drop ratio in this investigation . Experimental Result s Measurements of the pressure drop ratio for thin transvers e sections less than a fiber length in thickness of the heartwood of initiall y green western hemlock and air-dry white pine at different equilibriu m relative humidities are shown in table 2 . Air was passed around the outsid e of the sections and out through stopcock S! (fig . 1) for 2 days before passing it through the section in order to bring them to equilibr iu wit h the vapor pressure of the air . Pressure drop readings were then made ove r a period of 2 days . The total pressure drop used was never greater tha n 20 cm . of water . The data show that the fiber cavities change but ver y slightly in size with shrinking and swelling of the wood . The mean deviation of the effective capillary radius for the white pine was abou t 0 .5 percent and for the western hemlock specimens still less . In the cas e of the white pine a part of the permeability may have been through resi n ducts . These are entirely absent from western hemlock so that th e permeability of these sections was entirely through open fiber cavities . This approximate constancy 6f the size of the fiber cavities is i n agreement with findings of Schwalbe and Reiser (6) and Geiser (1),en d with the density-shrinkage relationships for small sections of wood drie d under as nearly stress-free conditions as possible developed by th e author (10) . j, R1075 -6- Table 2 .---Effect of cI .an,ses in relative vapor pressure upon the permeability of transverse sections of wood less than a fiber length i n thickness :Dimensions of : Capillary sections : constant Species : re : : Thick : Cross : c ness :section: --------- : -- -car . : : White pine heart . : 0 .22 :' Mean :Relative :Equilibrium : vapo r pressure : deviatio n dro p :pressure : ratio, : of air : -----Percen t ---------- - 1.37 : 1.85 x 10 0: 0.963 : .867 2.0)4 .505 2 .08 : : a .310 .117 .000 .117 .310 . . Western . hemlock : .heart .73)4 5 .30 1 .47 x . . . _ 5 .30 : 1 .47 x 10-7 : .117 .310 .505 4 734 .867 .963 I; : -7-- 2 .08 260 2 2.00 9 .50 9 .50 9,69 9 .53 9 . 63 2 .89 .73 )4 2 .8 6 .117 2 .93 2. 2 ..96 2 .98. 2 .9 4 . . . . . : .: : 2 .01 ', : 9.6 8 .963 .505 .73)4 .067 : 9,8 8 9 .62 9 .6T .000 .117 .310 810 75 . 1.95 .31 0 : 2 .05 2 .05 1 .99 .734 .505 .310 .117 .000 . .4o l0_7 : . : 2 .10 17.34 •.963 . 22 1 .96 2 .90 2 .89 : .94 1 .43 Similar measurements were made on initially green transvers e sections of both the sapwood and the heartwood of western hemlock greate r than the maximum fiber length in thickness . The results obtained fo r several relative vapor pressure change cycles are shown in figure 2 . Th e curves are numbered in order and the desorption part of the cycle indicate d with open circles and the adsorption part with black circles . There is a definite increase in the pressure drop ratio with ,decreasing equilibriumrelative vapor pressure indicating that the effective capillaries in the pi t membranes increase in size-'when the wood shrinks, During the first cycl e true equilibrium was not obtained or a slight checking of the sections ' occurred . After this the relative vapor pressure cycles were quit e case s reproducible . Adsorption values of the pressure drop ratio were in . These hysteresi s slightly greater than the corresponding desorption values e loops are entirely comparable with those for the moisture content-relativ vapor pressure (11) and electrical conductivity-relative vapor= pressur e The electrical conductivity-moisture content relation relationships ship calculated from the relationships of each with the relative vapor pressure . Hive a relationship free f ;com'hysteresis effects .' The same is true when th e pressure drop ratio-relative vapor pressure relationship is converted to a pressure drop ratio-moisture content relation 'Ship basis . The pressure dro p ratio is proportional to the fourth power of the effective radius (se e ovation 4) . The square root of the pressure drop ratio is hence proportiona l to the effective cross-sectional area of the capillaries . The relationshi p between the square root of the pressure drop -ratio and the equilibrium moistur e content is plotted in figure 3 for the sane data as in figure 2 . A linea r relationship free from hysteresis effects is obtained below' a moistur e content of approximately 20 percent . The external volumetric and crosssectiiongl swelling of wood is directly proportional to the moisture conten t from oven-dry to the fiber-saturation point (10) . .The cross-sectional area , of the pit membrane openings thus appears to be inversely proportional to th e extent of swelling of the membranes . Evidently - the thin pit membrane s respond to changes in'the relative vapor pressure more rapidly than doe s the relatively heavy rim about the membrane . In the course of desorptio n the membrane tends to shrink on drying . As the response of the heavy ri m is considerably slower the membrane is put under tension . This tension i s relieved by internal shrinkage which results in an increase in the cross section•of the openings . all (O. Similar determinations of the relationship between the moistur e content and the square root of the equilibrium pressure drop ratio were mad e on sections of both seasoned white pine he-tirtwood and initially green heart wood and sapwood of western hemlock cut in the three different structura l directions using the second apparatus . The measurements were made with total pressure drops ranging fro m '5 to 100 cm . of mercury for the different sections . Readings at at leas t three different pressures were taken with each section . The pressure drop R1075 Relative Vapor Pressur e Figure P .-Pffect of changes in the relative va p or Pressure of air p assin g through transverse sections of the heartwood and the sapwoo d of west e rn hemlock upon the resulting equilibrium pres s ur e dro p ratio . :!,''mr.bers indicate'or d er of measurements . () Desorption : Okisorption . I 1 16 18 20 22 Moisture Content (Percent ) Figure 3 .- Pffect of changes in the moistu r e content of transverse section s of the heartwood and the sapwood of western hemlock canon th_ e square root of the equilibrium uressure dr o p ratio . Numbers on curves ref e r to those of figure 2 . 0 Desorution ; (Adsorption . ratio Pc was plotted a : ;ainst the total pressure drop Pc plus Pw and the _eh _se value o Pc extrapolated to zero ap plied pressure was used in th e calculations . This graphically corrects for impact turbulence effects ( g ) . The pressure drop ratios built up to a maximum and then decreased to a n e quilibrium value which was used in these calculations . This effect fo r the vapor flow was considerably less than that reported by Buckman, Schmit z and . Gortner (2) for liquid flow,as might be expected for electrokineti c effect would undoubtedly be less for vapor flow than for liquid flow . If Sutherland, Johnston and Maass' (12) contention that the pi t membranes stretch under higher pressures, thus increasing the permeability , is correct, then the effect should be greater for the passage of air at hig h relative vapor pressures than for dry air because of the increased elasticit y of the pit membranes . The slope of the Pc versus Pc plus Pw lines wa s greater, however, at zero relative vapor Dress-are than at 90 percent relativ e vapor pressure . This is just the reverse of what it would be if the plasti c membranes stretched appreciably . The results obtained can, however, b e explained on the basis of the imp Act turbulence effect increasing at highe r velocities of flow . This will explain the deviations from the linear velocity of flow-pressure relationship obtained by these authors as wel l as the data of this paper . The results of these measurements with the second apparatus ar e given in figures Li, 5, and 6 . In all cases the square root of the pressur e drop ratio increases in a linear manner with a decrease in the moistur e content from about 20 percent to oven dry . Further, the ratio of the square roots of the pressure drop ratios at 0 percent and 20 percent moistur e content are practically constant for sections of different thickness and ' even for sections cut in the different structural directions, but the value s seem to vary with the species . The values of this ratio are given i n table 3 together with the thicknesses of the sections and the standar d pressure drop ratio for unit effective thickness . The measurements mad e by the first method are for a varying cross section caused by shrinking an d swelling of the section . The measurements made by the second method are . . for a fixed cross section determined by the size of the opening in th e flange . As the wood shrinks the number of capillaries that are effectiv e for flow increases . For this reason the ratios of the square roots of th e pressure drop ratios obtained by the second method are higher than thos e obtained by the first . The values obtained by the second method wer e corrected to the basis of the first by dividing by 1 plus the cross- sectional shrinkage from 20 to 0 percent moisture content (see last colum n of table 3) . These ratios of the square roots of the equilibrium pressure drop ratios seem to depend upon the nature of the membranes traversed rather tha n the arrangement and number traversed in series or parallel . If the pi t 81075 . -9- 1. 8 1 .6 Figure - .--Ffiect of changes in the moisture content of transverse sections o f the heartwood of white pine upon the square root of th e equilibrium pressure drop ratio . Desor .:tion ; Q Ansor ;tion . { Moisture Content (Percent ) Figure 5 .--Effect of changes in the moisture content of tangential and radia l sections of the heartwood cf e;hite ;.p ine upon the square root o f the equilibrium pressure drop ratio . LS Desor)tion ;C Adsorption . oTqug do,ac a .znsss .ld do 4cog aatenb g Figure 6 .--Effect of changes in the moisture content of transverse, tangential , and radial sections of both the heartwood and sap w ood of western hemlock upon the square root of the equilibrium pressure dro p ratio . All desorption curves . Ordinates as given ; >< lU ordinates ;() 100 '1, ordinates . Table 3 .----Changes ih permea=4lity of wood with changes in thickness and. moist - ire content Species : Kind of . section (1) • (2) Ratio of the . Standard. : : Thick- :pressure drop : square roots of : :Sec- : , ness ratio for : the pressure : :tion : : effective : drop ratios fot : of No . : sections : unit thica- : dry wood and : :+ wood at 20% . • . ness of ' dry wood_' ' :moisture content : (6) : (3) (4) (5) • -___ : __ _-; Do Do Do Do 1«300 ' : 5 .15 x 10 5 : : 1 .3WI : : 1 .857, : 2 .8e0 : : .184: : .189' : 5 .25 x 1J 1 .200 : 1 .26 x 10"80 : -: 1 .18 x 10_ : : 1 .1'4 x 10 -0 : 1 .25 2 1 .30 2 1 .26 8 1 .17 6 f 1 .193 1.180 . : 5 .57 x 10 - 13 : ; : 1 .79 x 10 - 13 : .1 .215 1 .230 1 .15 3 1 .16 8 .186 . :J) .50 x 10-12 : . .196 . 4 .58 x 10-1 3 ; .169 : 6 .29 x : : : : : 1 .20 0 1 .245 . 1 .222 1 .4'2 0 1 . 475 1.47 5 1 .32 2 1 .37 3 1 .373 : 2 .45 x 1o- 12 : : 1 .56 x 10" 12 : : 1 .03 x 10"' 12 : ' 1 .400 1 .385 . 1 .368 1 .33 0 1 .31 5 .112• : 1 .10 x 10- 13 : .140 : 5 .00 x .10-14 : '1 .365 1 .352 1 .335 2 : .158 .202 3 : .255 : 1 : 2' : ? 1,.23 3 1: :Heart . tangential : 1 : 1 .17 5 1 .17 5 :Hear t c 1 : transverse : 1 : 1 .14& - : 4 .13 x lo - -' : r,' : . . .do x 10`• 9 : 2 : 2 .12-5 ' : •3 .70 do 1.85 x 10"-' : 3 : 2.980 : . .' .do : . . . do :Heart : radial ' do (7 ) .M_-_ ; Western :Sad : 10 hemlock . : transverse : Do :Heart : 10 • 1transverse : Do :---- do : 1 Do : . . : .do 2 Do : . . . .do : 3 Do :Sap : . : tangential : 1 : . . . . do Do : 2 Do :Sap radial . . . . •: 1 Do :Hear t : tangential : Do :Hear t radial White pine Do ' Do Do Column (6 ) corrected. fo r externa l swelling 1 .32 2 was 3 .73 x 10 `9 ' -These measurements were made with the first apparatus ; and the effective cross section 1 .37 sq . cm ;. 1 .30 0 c . 4 r The other measurements were made with the second apparatus ;. - 2 was 5 ..27 x 1 0 lc and the effective cross section 0 .353 sq . cm . R1075 -10- . membranes which are composed of lignin have a similar hygroscopicity t o that of the wood as a whole, 'which seems to be the case, the volumetri c shrinkage of the membrane substance from 20 to 0 percent moisture conten t should be about 20 percent . If it is assumed that the rim of the membran e changes dimensions by a negligible amount wring the course of the shrinkag e of the membrane, hypothetical values for the ratio of the square roots o f the pressure drop ratios for 0 and 20 percent moisture content can b e calculated for various specific cases . If the shrinking manifests itsel f entirely as a change in the thickness of the membrane, the ratio would b e 1 .12 . If the shrinkage of the membrane manifests itself entirely b y increasing the size of the openings in the plane of the membrane, the valu e of the ratio would depend upon the fractional void cross section of th e membrane . If this were one-half, the ratio would be 1 .20, if one-quarter , the ratio would be 1 .40 . In reality . there is very likely a change in bot h the thickness of the membrane and in the size of the openings . If one-hal f of the shrinkage manifests itself in the direction of the thickness of th e membrane and the other half in the plane of the membrane and the void cros s section of the membrane is one-quarter, then the ratio of the square root s of the pressure drop ratios will be 1 .20, This hypothetical value is i n reasonable agreement with the actual experimental values . If the capillary structure of wood were quite uniform th e standard pressure drop ratios per unit of effective thickness should b e fairly constant for each structural direction . This is only approximatel y true in the fiber direction and not at all true in the other direction s (table 3) . All the measurements made on transverse sections were fo r sections only a few fiber lengths in thickness . The effective thicknes s of these sections is hence appreciably less than the actual thickness a s air passing through the sections traverses on the average one-quarter o f an open fiber length at each end . The number of pit membranes traverse d in series is hence approximately proportional to the thickness ef th e section minus one-half of the average fiber length (0 .17 era.) .. Thi s effective thickness was used in the calculations of the standard pressur e drop ratio per unit effective thickness for the transverse sections . The number of membranes traversed in series for the other sections was so grea t that the difference between actual and effective thickness was negligible . The standard pressure drop ratios per unit of effective thickness decreas e somewhat with increasing thickness even for the transverse sections . Thi s is due to the fact that the probability of maximum-sized openings occurrin g in series decreases with an increase in the number of membranes traversed i n series . It can be readily demonstrated that the permeability is greate r if the maximum-sized openings occur in series rather than when they occu r in parallel . The data show an appreciable deviation from the linear moistur e content-square root of the pressure dr op ratio relationship at moistur e contents exceeding 20 percent . The square roots of the pressure drop ratio s arc less than the linear relationship calls for . This is dub to the fac t that moisture condenses in part of the effective capillaries even belo w the equilibrium saturation pressure because of their minute size an d R1075 -11- eliminates them as a source of flow . The relationship between the curvature of a drop and its equilibrium vapor pressure is given by Kelvi n1 s equation x-- . 2 Cr M pRT In po p ., (5 ) in which r is the radius of the drop or of the capillary in which i t forms, po^is the saturation vapor pressure at the absolute temperature T2 p is thr vapor pressure over the drop or liquid in the capillary, R i s the gas constant, 0 -is the surface tension of the liquid, M the molecular weight, and %o the density, all expressed in centimeter-gram-seconds . The -relative vapor pressure in equilibrium with the moisture content at whic h the deviation from the linear relationship begins is practically 0 .9 fo r all of the curves . This corresponds to a capillary radius of 10 rn7 .. I t is slitly less than the average effective capillary radius previously n determined by other physical methods (a) . At moisture contents i equilibrium with higher and higher relative vapor pressures more and mor e of the capillary structure is`obstructed due to condensation of films o f water that cannot be broken at the pressures under which the measurement s were made . Measurements were also made of the pressure required to over come the effect of surface tension of water in the saturated section s using the second apparatus . This proved to be more sensitive to th e detection of the initial flow of air than the method previously used . For all the transverse sections, detectable displacement of water by ai r occurred at lower pressures than previously reported (8) . These pressure s in the case of the transverse white pine sections were less than the maximu m operating pressures used in making the measurements on the same sections a t lower moisture contents (15 to 20 cm . of mercury) . Finite values of th e square root of the pressure drop ratio were hence obtained at th e saturation moisture content of 29 percent (fig . 4) . In the case of th e transverse sections of western hemlock (fig . 6.) the pressures required t o overcome the effect of surface tension ranged from 0 .5 to 2 kg . per squar e centimeter . Although these values exceed the pressures used in th e measurements, the corresponding capillary sizes, 70 to 2 , g00 mp, are s o large that the vapor pressure depression bY them is less than 0 .1 percent . The square root of the pressure drop ratio will thus be zero at practicall y the saturation moisture content (30 .7 percent) . In the case of th e tangential and radial sections of both species the pressure required t o overcome the effect of surface tension ranged from 20 to 25 kg. per squar e centimeter, corresponding to capillary sizes of 75 to 60 mpl and relativ e vapor pressures of 90 .6 to 9S .2 percent . These correspond to moistur e contents of 27 .2 percent for the white pine and 2E .9 percent for th e western hemlock . R1075 --12 - r The distribution of size of the capillaries effective i n controlling the rate of flow of air can be calculated from the deviation s from the linear moisture content-square root of the pressure drop rati o relationship . The rate of change of the pressure drop ratio or the rat e of change of the velocity of flow with changes in the effective capillar y radius are plotted in figure 7 for a tangential section of white pine . The relative humidities in equilibrium with a number of different arbitrarily chosen moisture contents along the part of the curve whic h deviates from the linear relationship (fig . 5) were used to calculat e the capillary radii at which the formation of films across the capillarie s occurs . The fractional change in the pressure drop ratio divided by th e corresponding capillary radius change increment was plotted against th e average capillary radius for that increment . Sufficient experimenta l values for the pressure drop ratio at different equilibrium relative vapo r pressures were not obtained to determine the distribution curves with an y certainty but the example given gives the general nature of the curve . ! The most probable radius is 25 m .t ana the average 27 mil. This latter valu e is in good agreement with the average values previously obtained b y combining :electrical conductivity data and pressure permeability data ( ,2) . This method for obtaining the distribution of size of pore s should be applicable for all kinds of membranes in which the openings ar e sufficiently small to appreciably reduce the relative vapor pressure . I n the case of nonswelling membranes the linear part of the relationshi p would, of course, be parallel to the abscissa axis . Summary Measurements have been made of the equilibrium permeabilitie s of softwoods to air of different relative vapor pressure . Transvers e sections less than the average fiber length in thickness, in which th e open cavities account for practically all of the permeability, show practically no change in permeability with changes in the equilibriu m relative vapor pressure, thus indicating that the size of the fibe r cavities changes but slightly upon shrinking and swelling of the wood . Sections thicker than the maximum fiber length, the permeabilities of whic h are dependent upon the size of the pit openings, show an increase i n permeability with a decrease in equilibrium relative vapor pressure . The same is true for all tangential and radial sections . When the square roo t of the permeability is plotted against the moisture content of the wood i n equilibrium with the various relative vapor pressures of air, practicall y a linear relationship is obtained from 0 to 20 percent moisture content . The ratio of the square roots of the pormea bilities for 0 end 20 percent moistur e content are practically constant for sections of differen t thickness and sections cut in the different structural directions but diffe r R1075 -13- 1 A I cr.; 0 F-I r--l r-i •rl ( sr:77ra MU. 0 0 0 Ca SnTpV ,--c, P. "S''F- A 0 O TC-CCCIO .I(T q.53 0 re-) ap (T13) 0TIT03E Lre TTT d O aA T4 00 'T Se2UaI 7qTA JAOTJ ;O 1CqTOOT9A jO 0. ;g @1,1 4 GUrT.,10 04). Figure 7 .--Distribution of the size of t,- ..e capillaries effective i n controlling the flow of air in a . tangential sectio n of white pine (Tl) . for different species . At higher moisture contents the permeabilitie s are considerably less than the linear relationship calls for . This is due to films of water forming across the capillaries . Higher pressures than those used are recruired to overcome the effect of the surface tension o f the water in these capillaries . A new means of determining th e distribution of size of openings in a porous membrane based on these finding s is given . r Literature Cite d (1) Beiser, W . Kolloid Z . 65 :203 (1933) . (2) Buckman S . J ., Schmitz, H . :and Gortner, R . A. 39 :103 (1535) . J . Phys . Chem. (3) Duclaux, J .) and Errera, J . Rev . Gen . Colloides 2 :130 (192+) . (4) Hitchcock, D . I . J. Gen . Physiol . 9 :755 (1926) . (5) Johnston, H . W., and Maass, O . Can. J . Res . 3 :140 (1930) . (6) Schwalbe, Co G ., .and Beiser, W . Papierfabr. 50:655 (1933) . (7) Stamp; A . J- Colloid Symposium Monograph ( 1 927) ; 6 :83 (1928) 0 4:246 (1926) ; 5 :36 1 (8) Stamm, A. J.- J . Agr . Res . 38 :23 (1929) . (9) Stamm, A. J . Phys . Chem. 36:312 (1932) . (10) Stamm, A . J. Ind . Eng . Chem. (1935) . (11) Stamm, A. Je, and Loughborough, W . K. J . Phys . Chem. . 39 :133 (1935) . (12) Sutherland, J . H ., Johnston, H . W ., and Maass, 0 . Can. J . Res . 10 :36 (1934 ) . (13) Walker, A . C . J ._ Text . Inst . 24 :T145 (1933) .