Document gaRMba88qQjNko18GZ15dk9VN

o3 t : 202 B. F. Goodrich Chwftical Company A DMMM OF TH* H. F. QOOOWCW COMPANY BASS OP REMOVAL OP SSHOAL VHYL CHLCRHS racN auawsiai fvc By Warren A. Reed Date C< October 28, 1969 Date leaned: Hoveooer <*, Akron STlC. Bean P.T .Whitmire - R.S.Reynolds Ameripol J. S. B. Wolfe Avon Lake General Chemical P. P. Reber R.H.Rylands - R.W.McKay *J. M. Whitney Brecksville Research Center A. R. Berens R. J. Pavcett *C. F. Gibbs E. i. Set* G. X. Stneber Research Center Library D. E. Ley Cleveland O. F. Beckmeyer W. F. Bixby *W. S. Brodine E. R. Clayson A. M. Fairlie E. W. Harrington J.M.Hyslop - P.W.Shore M.W.Laraon - C.D.Segner (7) P. H. Lawrence *G. H. Metzger S. L. Paulsen G. D. Schaaf R.D.Scott - J.L.Nelson B.M.G.Zvicker - Jeanne Valentine Calvert City *C.L.Woods - D.B.Schrock Henry C.B.Cooper - J.P.Piers H. L. Kuchenmeister Long Beach A.W.Clements - W.D.Robb R. K. Miller Hiagara Falls T.R.Linak - H.R .Aquino Welland DJi.Brooks - J.J.Dunn Pedricktown J. W. Goetsch W. T. Kloepfer A.R.Webber - J.A.Klupar Louisville R. A. BaTT-tan H. E. Barrett M. J. Carmack L.G.Crunkleton - S.S.Michels W. Yesowitcb Development Center L. F. Arnold *D. J. Bamicki *L. A. Bennett R. R. Bloor H.L.Brandt-G.M.0'Dell E. A. Collins *L. H. Conklin *B. A. DiLiddo *C. E. Fleming *C. H. Kotheimer F. F vrause R. M. Kreager *C. H. Lufter H. H. Marty *R. J. Meyer R. S. Morgan *C. E. Parks *N. H. Sherwood J. A. TePas *G. L. Wheelock *R. J. Wolf F. W. Zieber Technical Council Central Technical <2) Modified Report BFG04561 20689001 TABIE OF CONTENTS Page SUMABY .......................................................................................................... CCHCUJSICBS ............................................................................................. RBCGMBHDATI01S ..................................................................................... niscussiai ................................................................................................ Experimental............................................................................... Results.................................................................. Vinyl Chloride Efficiency.................................................... AFOTDEC .................................................................................................... HEFEHE9CES ............................................................................................... 1 1 2 3 3 3 9 11 20 ZU06S30 BFG04562 1 SUMMARY It has long been recognized that physical expansion due to rapid vaporization of residual vinyl chloride aaooaer can affect the plasticizer absorption properties of FVC.U) This is one of the advantages of the use of blordown tanks. On the other hand, charge size, blowdown tank volume, turn around tine, conversion, recovery capacity, liquid level in the blowdown tank prior to dropping a charge, etc., all affect the efficiency of the blowdown/recovery operation and therefore affect the particle structure. Many "unexplained" shifts in resin properties or plant to plant differences nay be due, at least partially, to differences in recovery of residual monomer. Exploratory work reported earlier^ demonstrated that both conversion and rate of monomer removal from the particle influence the porosity of the resin. In the experiments reported here, we have attempted to better quantify this effect. Similar or related observations at various plants are cited. It is proposed that through the use of a thin film evaporator or cyclone we may be able to control the rate of removal from the particle better than we have in the past, make this step more uniform from plant to plant and markedly improve the efficiency of the vinyl chloride recovery process. CQKLUSIOHS 1. The rate of recovery of vinyl chloride monomer from suspension IVC affects the properties of the realm. a. Rapid recovery produces higher porosity, lower density and shorter powder six time at n given conversion. b. Rapid recovery msaa^Sjoduce faster powder mix time at a given oorealty. 2. Monomer recovery rate should he controlled in the manufacturing process to improve control over resin properties. 3. A continuous stripping operation should both improve control over resin properties and improve efficiency of vinyl chloride recovery. W.A .Reed to G.D.Schaaf, "Blowdown Tanks; Recovery Rate", January 9, 1968. 20683003 BFG04563 RECOMGNDATION S 1. Obtain process design data for continuous stripping of FVC slurry. R. A. Forsthoffer'and William Kloepfer have been working on a pilot stripping process at the Development Center. 2. Incorporate continuous stripping in the Louisville 10,000 gallon polymerizer process design. 3. All plants should review their hlowdown/recovery operations to be sure this is carried out as uniformly as possible from batch to batch. 2 R.s.Morgan to R.A.Forsthoffer, "Vinyl Chloride Recovery System Study Experimental Series". July 9, 1969. Discussion 3 Physical expansion due to rapid vaporization of residual monomer affects the plasticizer absorption properties of PVC. This is particularly true at low con version and has been recognized for many years. This principle is one basis for the use of blowdown tanksA5)jn an earlier report we again demonstrated that porosity is a function of both conversion and rate of monomer removal in suspen sion IYC.(l) Of course other factors such as choice and concentration of dispersant, water to monomer ratio, impurities, etc. can influence these rela tionships, also. The experiments reported here were run in an attempt to verify and quantify the effects of conversion and monomer recovery rate. Several plants have noted unusual results of process changes which may be due, at least in part, to differences in monomer recovery. In his work on the hydroful process M. J. Carmack has noted an effect of recovery rate on resin porosity.(2) At a PVC Task Force meeting R. K. Miller reported the following data demonstrating the magnitude of this effect on porosity of resin ssmples at Long Beach. Table 1 Porosity of EP Resin Sampled at Long Beach Sampled from Poylyl (IUMonomer flashed off at Beginning of ^ Qff Recovery J I Slowly Normal Charge 0.U2 ml/gm. 0.290 ml/gm. Charge Shortetopped with BBA O.hO al/gm. 0.265 al/gm. Sampled After Normal B< 0.275 ml/gm. 0.263 ml/gm. More recently, M. J. Carmack ham reported porosity and IMF equivalent to normal production at higher than normal conversion achieved by reducing charge size.() This shift is very likely due to more rapid recovery rate at the reduced charge size. Another question, which can be explained by the influence of blowdown/ recovery conditions, is the difference between the reproducibility of timed cycle hydrofill operations at Louisville. (9) ^ V.A.Reed to G.D.Schaaf, "Blowdown Tanka; Recovery Rate", January 9, 1968. M.J.Carmack to S.S.Michela, "Statua Report No. 8 - Hydrostatically Full Poly", May 1, 1967. M.J.Carmack to H.E.Barrett, "Reduce VC1 Losses", August 7, 1969. (9) W.Yesowitch to H.E.Barrett, "Statua Report - Hydroful vs. Time Blowdown Objective", September 19 1969- H.H.Marty, "Background Information on Blowdown Tanks vs. No BDT's", April 11, 1969. 206S9QO5 BFG04565 4 Carmack and Ballman^ have also pointed out a shift in the porosity vs. powder mix time relationship which could be related to recovery rate differences. 103EP F-73 102EP F-5 (L) A. M. Fairlie' ' also noted a similar relationship between BIT and porosity for normal production EP resins at ALGC. A shift in this relationship was noted for samples blown directly from the poly to a sample container. (3) M.J.Carmack and R.W.Ballman to R.K.Miller, "Conversion Control of VC1 Homopolymer Reactions Based on Premature Pressure Drop - Status Report Ho. 3 by R.K.Miller", March 1, 1968. (0 A.M.Fairlie to R.W.McKay, "Relationship Between Por< Tests for Geon 102EP F-5, May 18, 1967. i BfG04566 5 In both these cases rapid recovery not only produced higher porosity but faster powder mix at a given porosity. In an unpublished investigation, R.S.Morgan and W.A.Reed attempted to correlate HC and porosity results from AIC and noticed shifts in this relationship with tine. (This technique can be useful when attempt ing to identify process shifts and ultimately gain better control over the FVC suspension polymerization process.) Many other examples of differences between product made in different plants or under different conditions could be cited as evidence for a substantial effect of monomer recovery rate on the nature of the suspension IYC particle. However, we must now attempt to gain better control over this part of our process. Experimental These experiments were run in the Development Center 175-gallon glass-lined poly with off-center marine blade agitation. The batches were run hydroful measuring conversion by the weight of water added to compensate for shrinkage. At the desired conversion the charge was shortstopped Then approximately half the charge was blown rapidly to an evacuated 3300-gallon blewdown tank. Piping was arranged so that the slurry stream from the 175-gallon poly entered the blowdown tank near the top and directed tangentially to the inside of the tank so that the stream would swirl around the surface of the blowdown tank before falling into the liquid phase at the bottom. This was done to provide good surface exposure for evaporation of vinyl chloride. Then monener was slowly vented from the slurry remaining in the poly. (For additional details of procedure and recipe see the sample Manufacturing Specification attached in the Appendix.) Results Charges were run to 60, 75 and 9Oft conversion. Samples ware obtained for slew and rapid recovery. The effect q&these variables on porosity is shown in Figure 1. This verifies the trends showfein the previous work. Conversion effect is quantified. The range of recovery rates covers what might be expected at different locations but is not quantified and may vary somewhat from charge to charge. Data are presented in Table 1. The effect of these variables on compact density and powder mix time are shown In Figure 2. Of course any other variable affecting particle structure such as agitation, dispersant type and concentration, oxygen, etc. will shift these relationships. In addition to these relationships, Joseph Phillips ran porosity on samples fractionated on a particle size basis. In general, at low conversion, the coarser particles were more porous than the fine particles. At high conversion the spread in porosity was rather narrow. These data raise the question, "Are fine particles actually at higher conversion than the coarser particles?" $FG0456'7 Figure 1 Effect of Conversion and Recovery Rate on Resin Porosity 6 Jt CM > S2 H 31 HrXHI H r0C* -Cas--* cCoM i 1 oCMnON CCM'"* Ot--xUONCO COQv O O HO CrHO CHO VrHO COM CM CM On $ S UN .g 1 -C3MCoM OC*MN-Ca*"- MVNN*O^ Ut-N O ITS COM oonn CHM oH tH- ^ cvi cO ^ S\ NO e* on X W 6h O_n o 5 ^^ <CoVnI Qol OmNUoJn ON Oro?NOSrHH Oc-- r--ji m 8 ^ 3 CO O CNI S Cl/ON HrH NrHO -3 eg S g> 2 X 3 X Sjd* rH O4r !lCfM\ OON tO--UC*N-44*1 O rH 0O On H4 CM tU--N on IOrTHS -a- t-- u4op> a$ dK#< !cOovcOn oH\loOnHe>H J NO on UN H CNI H UoN\ mON Jt SI 1A CO IHA & CM <n Hs ^ wO ^ ^8 \Q Kr1 O t 3*3 uHnmt*-0cOoC^O oO' O UN VHO CmVI UHN C no* 0O1N * H UN CM Si r> ^ouno-ocm8n8un . e- -C3M HCVI nNOo t- -(t3\--iHlCTM\ ^ 1 -OO3N mNON NrroHH mrMrHHN* CmrCHOM rOCmHM * CM o on -n3 4 O &8 o . . On CCMM UCVNI -r3H NO CM rH * XOrrH-i HXi HX*0<90 U(NM*OOOOl UONVOOOtn' O OJ OrH CONI nO>l VrHO <rSH) 4Ol1 cn jC---,3cCMn *cmH- mvo o mMmoN voo . crHn cm h mm vCoM vO jC*"- oCMn 3;? !!oCt~i-v-r3O- NCOOOoOOnn H UN H CO oi H CNJ M) vO mm UFN- UN ONO_N CMCcMm vOCoM Sv hO VOVQmO vo_O r-H -3- on SCM cHm Hac 8 2* 3 x 33 !cCo-COONO-3O CCNNII 0- Ot"N o co urHn conncnriH oCN-I -t031- ^mirNrH -c3cm- -drrHH'Cv_rOHO_mrOH* O^m5* \0QCm_M ^VoO s Jh 3 ^ 2 X 2 XV om Xv xO S0_5 Sample Id e n tific a tio n -- cCl hX-> -a5 5 ruS* H-rt a rc-vi O -c3ni O CuMn frrH- Or-Hn uoHn CcOo jCt*" -O3 NOO 0O0 0rOH-3OH 0COM -(>HC P ou hCAhSQ,l ^^OCoOH(X-J(S>qIPduwUXVCHsBCPISVoU-orIiNhplmrNHOlaO.xIHH(jrSTUO^)H*hHOrS>flHoU>luBIhr~aN`S0Vj-2HtB|gC+Opl> 0B-hHr-N S,n5 3>21^3|J.^ sx xPPOj** +h" 3ZTS STOI^JBd 7 or co co oo b?G04569 20689009 Figure 2 Effect of Conversion and Recovery Rate on Compact Bulk Penalty and Powder Mix Time 8 I o to q s s o z 9 What does this imply for the suspension PVC manufacturing process? First, monomer recovery rate must be uniform from charge-to-charge, lot-to-lot, etc. if we are to achieve proper quality control over the resin properties. If ve expect to make the sssk product in different plants or different buildings vithin a plant, recovery rate is one of the process variables which must be controlled. Second, in the normal "blcwdcwn" process the first slurry dropped into the blowdown tank is subjected to more rapid vaporization of residual vinyl chloride than the last. This could lead to non-uniformity of particle structure vithin a charge. I have proposed^ that the slurry pass from the poly through an expansion nozzle (for maximum rate of pressure drop), into a film evaporator (for separation of vapor from the slurry) and directly into the slurry storage or blend tank. This takes advantage of maximum rate of recovery going from polymerization pressure of about 100+ psig to the vacuus achieved by the low pressure recovery compressor. B. R. Clayson(^) and others from Cleveland Engineering have discussed this method of recovery with us and he has prepared a preliminary design for such a flash recovery system to be considered for use with the 10,000-gallon polys at Louisville. Even though all the monomer goes through the compressor, in this proposal, the size of the compressor is reduced compared to our normal design because the demand is more constant rather than cyclic. The only problem is the relatively long time required to recover the entire charge. Based on subsequent discussions about this process, I would like to propose an alternate design in which two stripper-separators are used. (See Figure 3). The first is connected to a high pressure recovery system. Slurry is transferred from the poly, mixed with steam and fed to a cyclone or some other thin film stripper-separator. This may be connected or lead to the bold up tank for feed to tbe second stripper. Vinyl chloride from this second stage then goes to the lew pressure recovery system. Shis second stage stripper is maintained at a constant low pressure for maximum monomer recovery. Temperature is controlled by addition of steam to the feed stream. Vinyl Chloride Efficiency Some type of continuous thin film stripping operation is necessary to improve the efficiency of vinyl chloride recovery operation. In lew pressure recovery from polys or -blowdown tanks, only the layer of material at the liquid surface Is sub jected to maximm driving force for evaporation of monomer. In a continuous thin film evaporator, tbe entire patch is subjected to the same conditions with no short circuiting. ^ WJV.Reed to G.D.Schaaf, "Blowdown Tanks; Recovery Rate", January 9, 1968. Earle R. Clayson to H.HJiarty, May 20, 1969. 06i t 8308 10 r 11 appendix BFG045'73 W I MNUFACTUONG SPECIFICATION NmN t 4'f* M W * * B.F.Goodrlcb Chemical Company t gililign 01 IHI I f CBOOIICH 12 Page 1 of 2 COMMOO'Ti s*Ck------ 0110-2740 _____ _ COWMCftCKU 9CI'CN*fiW.. Cnr^CAk Ow****'AT,*--. . 1. 0. T3.IE.. --P-~H--y--dr-o--s--t-a--t-i-c--a--ll-y--P u l--l ------ Polymerization OHOCtll___ Recipe Vinyl Chloride Methocel 60 HG 50 cp. Elv&nol 50-42 IPP (in l40 gms. Hexane) DJ4. Water (120 gal.) Bisphenol A lllUC HO . JtiL Weight w 68.1 gms. 34.0 gms. 34.0 gms. 1003# 1135 gms. itt -flay-A-j&a------- Psrts 100 0.030 0.015 0.015 200.6 0.05 Temoerature: 57*C Agitation: 350 RPM (Two 12" Marines at 14" and 30") Blowdown: See Item 10 Procedure ' 1. Prepare premixed sope solution: (a.) Add Elvanol to 1 gal. cold D.M. water* (4.) Heat to 80*C. (0.) Add Methocel (d.) Cool to 40*C or lower. 2. Charge 995# of D.M. water. Heat to 65*C Evacuate Hold 5 minutes. Break vacuum with nitrogen and cool to 40*C charge sope solution. 3. Evacuate poly, charge vinyl chloride* heat to 57*C and control. | 4. Predissolve IPP in 140 gms. Hexane. 5. Hook up two 20 gal bombs filled with D.M. water to poly. (Make sure there is a check valve in line between poly and beobs. 6. When control is reached (57*C) pressure catalyst solution into poly with nitrogen, fressure bombs with nitrogen to 30|E*i higher than poly pressure. Open valves between boob and poly. Record weight of Mratr added to poly. Take readings every 15 minutes for first two hours* then take reanxngs every 1/2 hour. 7 After hours* plot pounds water vs. time. Find intercept (A) which is pounds water to fill poly at zero time. Q, Continue to plot pounds water vs. time. When A+ 137-5 14s. water has 4een added} stop water addition. 9, Add shortstop dissolved in one quart of Methanol when A+ 161 lbs. water would have been REASON FOR ISSUE: ONidiNarco Of W. H. Scott Development Center Pl*P(T MANAqCN PLANT APPROVALS I CLEVELAND OFFICE iPAODUCT IM*AOvW(NT prOjUCTiOpa ma'.acCR r(CN<C4l M4NACC* NNOOUCTION ; technical DEVELOPMENT CENTER GO^14 B? MNUFACTURMN6N4SNPI ECCt>IFINICWAT*IOA N O.F.Goodrich Chemical Company 4 giniiOi Of CIXiMODI t * NO . ... I h t i f (OOuncH tonfiti 0110-27-0 -- ________ --____ ejMMtlCltti DO s 4 t.Q(, ch(.*mcw rn ij**TiCH. 103P-Uydro3tatically Full 13 Page 2 of 2 110X271 *ac c i PolyserlzatIon______ Procedure (cont'd.) > nq.__________ .. mm no 7~H_____mn . May 2, 19^9 added if water addition had been continued. *10. Ten minutes after shortstopping, blow one-hal? of charge rapidly to evacuated 33OO gallons wash tank. Hold vacuum on tank to dissipate excess vinyl chloride.* Vent and cool, one-half of charge in poly. 11. Centrifuge each half (Sharpies) have 1 lb. sample of each for lab. Tray dry at 50*C to less than 1% heat loss. 12. Screen thru 10 mesh and package (save tailings). 13. Becord net weight of product and tailings. T06S902 045^6 b?g 1 )6 8 9 0 2 hO! BFG04577 Micron* 0 e ! O bvg451& 0.01 0.0$ 0.1 0.2 OS 1 2 S 10 20 30 40 SO C< 70 80 80 9$ $8 99 99.5 99 5 19 9 Micron* BFG04579 I 0.05 0.1 0.2 0.5 1 * S 10 to 90 40 00 TO 00 0 0 OS 08 00 00.5 *9.8 18.9 M .M % Through 110DC271* iPOLT) 20689013 BFG04580 r Micron* 0.01 0.05 01 O.J 0.1 I ? SM M ) 0 40 SO to 70 Vo to ts #0 * >.S *9 * 9 M -H I B. F. Goodrich Chemical Company 16 Inter-Organization Correspondence T W. A. Reed Development Center Date Pram June 23, 1969 Joseph Phillips Porosity of Particle Size Fractions Procedure Sample 110X274 H-1125 (Wash Tank) (60f Conversion Shortstopped) 1. Sample was screened and the fractions collected. 2. The 420, 250 and 177 aileron fractions were combined (little or no 420 and 250). 3. Vote that the fractions are what vas retained on the screens. 4. Porosities vere determined on the original resin (unscreened) and each fraction. 5. The large size fraction (420*177 micron) under microscope shoved 60/40 particles with and without spikes. That la, the spiked particles reminded one of a sea "mine" with spiked detonators. The "spikes" vere. really particles in the range of 149 micron and less which vere fused to the larger particles. Vote: Under about 12CK these large spiked particles are really smaller fused agglomerates and appear to form a sintered mass. Particle Size (Microns) i On p*20 # <250 U77 149 105 74 < 74 .00 .69 5.88 10.38 28.03 34.60 20.41 nrPHta^ Porosity of tag.Ha e Size Ebroeity prigInal Tractions Orig Fract. faction Practice .43 :S 1 .806 1.75 .053 .12 .723 .833 .747 .653 .483 .43 .60 .091 .50 .209 .43 .226 .35 .098 .677 .06 .14 .15 .07 .54 6. The fractions below 177 micron were uniform in shape and appearance and progressively smaller. 1 0Z06890Z irg.iits) V. A. Reed -2- 110X271* H-1125 Poly Separated Particles 17 June 23, 1969 110X27** H-1127 Wash Tank (75^_ Sgl Separated Particlea BFG04582 N  00 20G8S0Z if. A. Reed 18 June 23, 1969 UOQC27U H-1128 Wash Tank Particle Size (Microns) ot 420 250 177 149 105 88 74 < 74 .00 .25 17.09 41.07 28.57 9.44 3-57 Porosity Porosity 1 1 1 pore Size Original of Fractions Orig. Rract Fraction Fract. .106 .106 I V J .114 .110 .106 .104 .088 .060 .26 .28 .27 .26 .26 .23 .19 .26 .020 .045 .030 .008 .002 .105 .111 .074 .022 .007 .262 cl Attacbmt. (l) cc: File- (3) c: 4 . -- Joseph Phillips BFG04583 *s; 110X274 (H-1125, H-1127, H-1128 19 Particle Size (Microns) Code $ 6oj Conversion Rapid Recovery - Porosity Range - .483 - .833 Avg. Porosity .658 0 6oj Conversion Slow Recovery - Porosity Range .275 - .338 Avg. Porosity .306 B Conversion Rapid Recovery - Porosity Range * .117 - .289 Avg. Porosity a .203 0 90& Conversion Rapid Recovery - Porosity Range * .060 - .114 Avg. Porosity * .087 By: Joseph Phillips Date: June 23, 1969 28338 BFG04584 E20S890Z REFERENCES 20 1. W .A .Reed to G.D.Schaaf, "Blowdown Tanks; Recovery Rate", January 9, 1968. 2. M.J.Carmack to S.S.Michels, "Status Report No. 8 - Hydrostatically Pull Poly", May 1, 1967. 3. M.J.Carmack and R.tf.Ballman to R.K.Miller, "Conversion Control of VC1 Hamopolymer Reactions Based on Premature Pressure Drop - Status Report No. 3 toy R.K.Miller", March 1, 1968. J*. A.M.Pairlie to R.W.McKay, "Relationship Between Porosity and Plasticorder Tests for Geon 102EP P-5, May l8, 1967. 5. H.HKarty, "Background Information on Blowdown Tanks vs. No BET's", April 11, 1969. 6. Earle R. Clayson to H.H.Marty, May 20, 1969. 7. R.S.Morgan to Kv^.Porsthoffer, "Vinyl Chloride Recovery System Study Proposed ExperMMtal Series". July 9 1969- 8. M.J.Carmack to H.E.Barrett, "Reduce VC1 Losses", August 7, 1969. 9. W.Yescwitch to H.E .Barrett, "Status Report - Hydroful vs. Time Blowdown Objective", September 19, 1969. *2 0 6 8 9 0 2 &FG04585 1 B. F.GOODRICH CHEMICAL COMPANY DEVELOPMENT CENTER AVON LAKE, OHIO Project 3 060 Does Stabilizer Reduce Effectiveness of PVC as Estane Flame Retarder? 14 November 1969 ELG 10-69 Introduction; There has been some speculation that a PVC stabilizer, inhibiting the evolution of HC1, might at the same time reduce the effectiveness of PVC as a component of flame-retardant systems for Estane. Conclusion: Based upon a simple Bunsen burner screening test, there was no significant difference in flame-retardance between the recipes with and without stabilizer. Procedure: V srf<jfc Equivalent Estane/PVC recipes one with and the other without stabilizer S-55" were compounded, milled and cut into . 075" X . 750" X 5" test strips The test strips were submitted to the Bunsen burner screening test. (1) Results: The recipes with and without stabilizer were equivalently classified as " non-burning ". Recipes and Teat Procedure Estane 5707 102EP F-5 P-2 F-19 L-2S L-27 S-55 100 5 5 5 .5 .5 .25 100 5 5 5 .5 .5 Test Procedure Test specimens . 075" X . 75" X 5" were held in a horizontol position and tilted 45 in a lab stand within a lab hood. A gas flame was applied to the end of the specimen for 15 seconds and pulled away to observe whether the specimen would continue burning. The flame was applied, within 15 seconds, as above for a second and third time. In each Instance burning stopped when the flame was removed from the specimen. *3 O 28731 o :o EGQ45gl T