EXPERIMENTAL SOUND~INSULATING IN KIN-SS-WOOD WARREN '8 1 BLOCK

Transcription

1 RESEARCH DEPARTMENT EXPERIMENTAL SOUNDINSULATING IN KIN-SS-WOOD WARREN '8 1 BLOCK WALL Reprt N. B066 ( 1958/8) hy9,..c!clk. - A.N. Burd, B.Sc., Grad. Inst. P. (W. Prctr Wllsn)

2 This Reprt is the prperty f the British Bradcasting Crpratin and may nt be reprduced in any frm withut the written permissin f the Crpratin.

3 Reprt N. B-066 EXPERIMENTAL SOUND-INSULATING WALL IN KINGSWOOD WARREN 'Bi BLOCK Sectin Title Page 1 INTRODUCTION 1 2 EXPERIMENTAL IDRK 2.1. Measurements n a Partitin f Small.Area Cnstructin aa a.iltbil8" Measurements.. a.aaaaail3l1:01, Measurements n Partitin Wall aa88a.j3. I Cnstructin..aaaaa&.c..e s Measurements by Cnventinal Methds a".l Crrelatin Measurements ea 8 0S Shrt Pulse Measurements 1l!l."O 9 3 COINCIDENCE EFFECTS 11 4 CONCLUSIONS 14 5 ACKNOWLEDGMENTS 15 6 REFERENCES I a a s J 9 a a i'i 0 all 0 15

4 April 1958 Reprt N ( 1958/8) EXPERIMENTAL SOUND-INSULATING WALL IN KINGSWOOD WARREN 'B' BLOCK SUMMARY The United States Natinal Bureau f Standards claims very high values f sund insulatin fr a type f partitin nt s far used in the B.B.C. Measurements made n such a partitin substantiate these claims. The cnstructin may be very useful when lw frequency insulatin is imprtant and flr lading is limited. 1. INTRODUCTION In a recent reprt f the Natinal Bureau f Standards 1 details are given f an interesting structure which fr its weight has apparently very high values f sund insulatin. It cnsists f tw expanded metal lath and plaster leaves spaced by spring clips frm a central framewrk. Table 1 shws the Natinal Bureau f Standards V figures f Sund Reductin Index (S.R.I.) fr the partitin and, fr cmparisn, results by the same authrity fr an unplastered 12 in. (30 cm) brick wall with a mass per unit area mre than six times as great. TABLE 1 Sund Reductin Indices f the Experimental Partitin cmpared with a 12 in. (30 cm) Brick Wall Frequency Experimental Structure 12 in. (30 cm) Brick Wall cls 19 lbs/sq ft 120 lbs/sq ft (93 kil/sq metre) (586 kil/sq metre) Sund Reductin Index (db) Sund Reductin Index (db) Average ' ( c/s) 55 53

5 2 As the reprt gives n indicatin f the reasns fr such high values, it was decided t make an investigatin n partitins f similar cnstructin. 2. EXPERIMENTAL WORK 2.1. Measurements n a Partitin f Small Area Cnstructin As a first attempt t determine the sund insulatin f such a structure, a dr t fit the entrance f the reverberatin rm at Nightingale Square was made. A timber frame cnstructed as shwn in Fig. 1 carried hrizntal crss-members spaced at 16 in. (41 cm) between centres, n each side f which were nailed's' shaped springs, (Fig. 2). Vertical i in. {O'63 cm} metal rds were tied with wire t the springs and a sheet f expanded metal lath was similarly attached t the metal rds and stapled at its edges t the wden frame. The dr was then supprted hrizntally and in. (1'9 cm) f sanded gypsum plaster was applied t each side frm underneath, this precautin being taken t prevent plaster intruding t far int the cavity and bridging the springs. = PIn view = \) (/ Fig. I - Dr frame fr measurements n partitin f small area = = = Side view Fig. 2 - Springs used in the cnstruct in f the experimental wall Measurements Initial measurements n this dr were nt satisfactry. The dimensins f the finished dr were insufficiently accurate and the frame int which it was fitted was fund t taper frm tp t bttm, a defect which had neither been freseen nr allwed fr. A layer f dense packing felt arund the dr sealed it int the pening but the sund

6 3 s /' ID ::> "... c ::> III / ,/,/ -.x. \ x"" )X "'- x >"J? (Y"'".J x " /' / /....//X'./ / x r-- / 00;000 V't 0... c:dqo '" et 000 Cl Cl' "" ""'d)ct\3- Frequency,c /s N :;:.. :"2" )('- -)(--)(' Fig. 3 - Experimental wall: sund reductin indices Results fr partitin f small area Hmgeneus partitin f same mass/unit area Blck wd dr (2i in,(6'3 cm) thick; 12'5 Ib/ft 2 (61 kil/sq metre» insulatin (Fig. 3) was nevertheless lwer than expected. Fr cmparisn the S.R.I. curves are shwn fr a typical 2 in. (6'3 cm) blckwd studi dr and fr a hmgeneus partitin f the same mass per unit area. Subsequent examinatin shwed that in spite f the precautins described abve, sufficient plaster had been frced thrugh the expanded metal t impede the actin f the springs and t cnnect the plaster surface t the hrizntal crss pieces in many places. Furthermre, since the expanded metal was rigidly fastened t the frame it culd nt vibrate freely n the springs and the timber framing acted as a bridge t transmit vibratins t the uter surface. It was realised als, after the cnstructin f the dr had started, that the high insulatin claimed by the Natinal Bureau f Standards might depend n radiatin suppressin phenmena, which are explained belw; tests n a small area partitin are cnsequently nt valid. The c-peratin f Building Department was therefre btained in a search fr a site where a full-scale partitin culd be erected Measurements n Partitin Wall Cnstructin A suitable site fr a full scale test was fund in the new extensin n the rf f Kingswd Warren 'B' Blck where a space was required t be partitined t allw cnstructin f a wrkshp n ne side and an ffice n the ther. A brick pier was therefre cnstructed, enclsing existing cable trays and water pipes running alng ne side wall, t leave a clear rectangular pening 16 ft x 10 ft (4'9 m x 3'05 m) fr the experimental partitin. In rder t prevent

7 4 Fig. J+ - View f part finished wall shwing springs, rds and scrim rigid fastening f any part f the partitin t the walls, thereb,y intrducing additinal flanking paths fr sund transmissin, a in. (1 9 cm) layer f crk was laid n the flr as a fting fr the partitin and in. (1 3 cm) Mafatex sftbard frmed the jint between the partitin and the walls and ceiling. Vertical timbers f 4 in. x 1 in. (10 cm x 2 5 cm) nminal sectin were erected, 16 in. (41 cm) between centres, in a frame fastened in this pening. Springs, as shwn in Fig. 2, were nailed t this frame at 16 in. (41 cm) between centres, hrizntal i in. (0 63 cm) metal rds were wired t the springs and scrim was stretched acrss and tied t the metal rds, as shwn in Fig. 4, t prevent plaster being frced thrugh the expanded metal n t the springs and the timber framing. The phtgraph was taken after plastering f the further leaf and shws that the scrim has prved effective. Expanded metal was then wired t the rds and plastered uing sand mixed with equal vlumes f lime and cement bund with cw-hair. The first attempt failed because the weight f the wet plaster pulled the springs utwards and dwnwards causing the wall t sag. T prevent this frm happening n the secnd side, vertical strips f plaster abut 6 in. (15 cm) wide and 2 ft (60 cm) apart were first applied and allwed t dry, s increasing the vertical rigidity. Temprary screws were put thrugh these strips int the timber and remved after the first cat had been

8 5 Fig. 5 - View f part finished wall shwing additinal vertical rds. The buildingbard insert between tp f frame and ceil ing may als be seen

9 6 cmpleted and allwed t dry. A secnd cat f sanded plaster was then applied and finally a setting cat f Sirapite mixed with lime. The thickness f the plaster was just ver in. (2 cm) except at the edges where it was bevelled t finish n the crk and sftbard surrunds. The first side was then stripped and replastered. T prvide additinal supprt, vertical rds were wired n t the framewrk and allwed t rest n the crk base (Fig. 5). In future cnstructins it wuld be an advantage t build the central framewrk with the timbers hrizntal and the metal rds vertical, standing n the crk fting t give additinal supprt. The precautins described abve t prevent sagging during plastering wuld then prbably be unnecessary Measurements by Cnventinal Methds Measurements f the S. R. 1. f the ",all were carried ut by the cnventinal methd f determining n bth sides f the wall the sund pressure levels prduced by warble tne frm a ludspeaker in ne rm. Measurements were made at quart.er ctave intervals between 50 cls and 250 cls and at half ctave intervals abve 250 cls, using ctave bandwidth filters t exclude nise and harmnics. Graphs are given in Fig. 6 f the results btained when nly ne surface f the wall was plastered, and als fr the cmpleted structure. IQ s "" " c... u '", i: "" c '" III ZO 10 v ) 1'1,,' I'X I...- I-<,/ R V.-.-e' ---,/,...-1\''/9" V',,,,J, /'/ y x x'xi x V 1-' -1-1, vi-',,'- V y. """ - - I-' v,,v /".-,.-, v It'", i--", " '-... ' :',/ vi,." I V - -- V I1, X \J / /"...» I/ V C> ", 0 Frequency, c/s,.; Fig. 6 - Experimental wall: Results f standard measurements n partitin wall l(--l(--)( One leaf f wall cmplete Bth leaves f wall cmplete Plane waves incident nrmally n a hmgeneus structure f the same mass/unit area Randm incidence results fr a hmgeneus structure f the same mass/unit area N.B.S. results fr this type f cnstructin

10 7 Fr cmparisn the N.B.S. results are shwn, tgether with theretical plane wave mass law figures and measured randm incidence figures fr hmgeneus cnstructins f the same mass per unit area. Observatins during the measurements suggested that much f the transmissin might be due t flanking paths such as the adjining crridr, the rf space and pssibly windws. Other methds f measurement designed t eliminate these effects were therefre tried Crrelatin Measurements The bject f this measurement was t identify and measure the sund reaching a micrphne frm a ludspeaker by different specific paths. Fig. 7 is a blck schematic f the arrangement used. In the arrangement shwn in Fig. 7(a) the utputs f tw micrphnes are fed int tw channels, in ne f which is a fixed time delay and in the ther a variable time delay. They are subsequently recmbined in a crrelatr built by Acustics Sectin fllwing the same general lines as that described by Gff2. It prduces a recrd f the crsscrrelatin cefficient between tw signals as a functin f the delay applied t ne signal. There will be a maximum crrelatin cefficient fr a delay crrespnding t the time f transit f sund between the tw pints frm which the signals are taken. Fir st psitin ". I_..' I ) I I I ) I ) (0) Prt it in First psitin, '-' Partitin Fig. 7 - Blck schematic f crrelatr measurements (a) Crrelatin f tw micrphne utputs (b) Crrelatin f ludspeaker input and micrphne utput In practice, it was fund that this arrangement had certain disadvantages and the alternative arrangement shwn in Fig. 7(b) was used, the input t the ludspeaker being crrelated with the utput frm a micrphne placed in turn n each side f the partitin. Bands f nise were used as the signal, the bandwidth being determined by cnsideratins described verleaf.

11 8 Fig. 8 illustrates (0) Wide band f nise the effect f signal bandwidth n the crrelatin diagram fr c 0 a typical sund insulatin.. measurement. Fig. 8(a) u C :::l represents a measurement using a very wide band f nise. -C 0.. The crss-crrelatin functin shws a single narrw peak at I» L. a time crrespnding t the L. 0 v 5 10 transit time f each pssible J Ul transmissin path. The height Ul 0 (b) Narrw band f nise f the peak is prprtinal t L. U the lgarithm f the pressure amplitude f the sund reaching the micrphne by that'path t and the difference between the peak heights in tw psitins will indicate the transmissin lss between them. Thus in 5 10 Fig. 8(a) the first peak at Delay, m.secs 3 ms represents direct trans- Fig. 8 - Effect f signal bandwidth n crss- missin thrugh the wall, and crrelatin functin the thers represent the (di agrammatic sketch) varius flanking paths. That at 10 ms clearly has a lwer transmissin lss than the direct path, Fig. 8(b) is the crrespnding diagram fr a narrw band f nise, say f nly ne-third ctave bandwidth. Each peak has subsidiary peaks f diminishing amplitude n either side f it, and it will be seen that the varius flanking paths are n lnger s easily distinguishable. It is therefre necessary t chse a bandwidth which effects a suitable cmprmise between path length discriminatin n the ne hand and frequency discriminatin n the ther. In the measurements n the Kingswd partitin a micrphne placed directly in frnt f the ludspeaker gave a grup f peaks at a time delay crrespnding t that intrduced by the spacing between micrphne and ludspeaker. An additinal delay was intrduced by the Pst Office lines between the experimental site and the crrelatr which, nt being readily prtable, was perated at Nightingale Square. As will be seen frm Fig. 9, the ttal delay due t all causes amunts t 4 t 5 ms. The micrphne was then taken t the ther side f the wall, the gain f the micrphne amplifier was increased and a new set f traces was pltted. N clear set f peaks culd be seen at the expected time delay and the S.R.I. was calculated fr the peaks f nise -the values being given in Table 2. The S.R, I. f the partitin must certainly be equal t r greater than these figures. Frequency Limits f Band TABLE 2 Results f Crrelatin Measurements 50 c/s c/s 1 kc/s - 6'8 kc/s 2'2 kc/s - 3'0 kc/s Sund Reductin Index >,. 52 db 68 db >,. 53 db

12 9 t c :;; u c.; c " t u t c u c.; c " " I 2 3 j 1 " m. secs Ca) I J -"" J1I 11l Uill -.L1 na A J J1.JL.1I /1 V \./"\1 \A1'li111 \ 'v IIJ vv v, y YI V IY Y I '" U ( I 2 3 " IS m. secs Cb) A,.j v'''.f>....m./i.., \M,)\ IIIAfi. fr\ A.)'W W... va.}f \f\nw'jii" c Frequency band:- 990 c/s-6'skc/s Fig. 9 - Examples f crrelatr traces (a) Trace with micrphne in frnt f wall (mic. amp. gain = 80- db) (b) Trace with micrphne behind wall (mic.amp.gain = 80 db) It will be nted that the S.R.I. in the narrw band 2'2 kc/s t 3'0 kc/s is 15 db lwer than that in the brad band 1 kc/s t 6'8 kc/s This is prbably due t flanking transmissin; the utput frm the ludspeaker was increased while using a restricted bandwidth and the "nise level" wuld be expected t increase by the same amunt if a selective flanking path existed at this frequency. The limiting S.R.I. which it is pssible t measure is thus reduced. N crrelatin peaks appear abve the nise, and the values quted in Table 2 are minima Shrt Pulse Measurements Fllwing the methd described by Raes 3, measurements were als made by applying pulses f a few millisecnds' duratin t a ludspeaker and measuring the amplitude f the first sund pulse received n each side f the wall. The amplitudes f the first three cycles reaching a micrphne in frnt f the wall were measured; this was repeated n the ther side f the wall after making the necessary adjustments t the micrphne amplifier gain. The transmissin lss fr each frequency was calculated frm: Al T.L, = 20 log (G 2 -G 1 ) + K db A2 where G1 = G2 = Al A2 K is gain in psitin 1 gain in psitin 2 average amplitude f 1st 3 cycles f incident pulse average amplitude f 1st 3 cycles f transmitted pulse the difference between the sensitivities f the tw micrphnes.

13 10 TABLE 3 Results f Shrt Pulse Measurements Frequency S.R.r. 125 cls 41 db 175 cls 50 db 250 cls 52 db 350 cls 40 db 500 cls 42 db 700 cls 46 db 1 kc/s 49 db 2 kc/s 51 db 4 kc/s 61 db In the analysis f the results, allwance must be made fr the effect f sund reflected n t the frnt micrphne. This micrphne was 6 in. (15 cm) rm the wall and the reflected pulse appeared at a time abut 1 ms after the beginning f the incident pulse. Measurements n the first three cycles fr frequencies less than 2'8 kc/s thus include sme reflected sund and an allwance shuld be made fr this III !: 40 J: U :: J: ::0 I/) f 3 0/ 20,,,l/' - / \ / /' /' I I.7." ,;,,-' \--- ;."" V"'".. /' X V ",/' ';'...;" -,V'...-- s-.: , "" /' C 10 '" )I").. 1I'I\C),...cD.9- Frequency,c/s '" C!. N Fig 10 - Experimental wall: Results f measutement n partitin wall c--c--c Crrelatr measurements } nrmal incidence Shrt pulse measurements A---A---A Plane waves incident nrmally n a hmgeneus structure f the same mass/unit area N.B.S. results

14 11 factr, depending- n the relative phases f incident and reflected sund. A crrectin varying- frm 6 db fr the lw frequencies t n crrectin at the hig-h frequencies has been included in the abve results. It is nt g-eneral practice in the B.B.C, t measure the sund pressure level at the surface f the wall but rather t take an averag-e ver a plane abut 18 in. (46 cm) frm the wall. Fig-. 10 shws results frm shrt pulse and crrelatr measurements cmpared with the N.B.S. results and als includes calculated fig-ures fr plane waves at nrmal incidence. 3. COINCIDENCE EFFECTS The results frm the cr relatr and shrt pulse measurements were fr sund at nrmal incidence nly. It was realised that n indicatin wuld therefre be g-iven f lwer values which mig-ht exist at ther ang-ies f incidence and tend t reduce the S.R.I. as nrmally defined and used. Such an effect has been pstulated by Cremer 4 and fund experimentally in scale mdel experiments by Schch and Feher 5 When plane sund waves are incident n a wall, flexural waves will be excited and travel alng- it at a speed which is dependent n the frequency. The incident wavefrnts in the air will als travel alng- at the same time, at a speed which is g-reater than the velcity f sund in air by a factr equal t the csecant f the ang-ie f incidence. Cremer shwed that at the frequency fr which the speed f the flexural waves is equal t the speed f prg-ressin f the incident wave frnt, resnance ccurs and there will be a vanishing-iy lw S.R.1. Thus abve a certain critical frequency there is a s-called cincidence frequency fr each ang-ie f incidence. The lwest cincidence - the critical frequency crrespnds t sund incident at a g-ra3ing- ang-ie upn the wall. At the critical frequency the velcity f the flexural waves in the wall will be equal t the velcity f sund in air (Fig-. 11(a)). Fr hig-her frequencies the velcity f the flexural waves is g-reater and the sund must be incident at sme ang-ie e such that the velcity f prg-ressin f the wavefrnts alng- the wall (c/sine) is equal t the flexural wave velcity (Fig-. 11(b)). Abve the critical frequency there is a reduced S.R.I. fr randm incidence, the decrease depending- n the si3e and the internal damping- f the material f the wall. These effects have been cnfirmed experimentally in the labratry using- a 3/16 in. (0'5 cm) sheet f aluminium which was sealed int a windw frame. If a diffuse sund field is set up n ne side the radiatin int free space shuld shw maxima in directins making- the cincidence ang-ie with the nrmal t the sheet. Warble tne at 4 kc/s frm a small hrn ludspeaker prvided the incident sund and the sund pressure level was measured rund the periphery f a semicircle. Varius features f the building- limited the arc available t that shwn in Fig Since the surce was f finite width it was necessary t make measurements at a cnsiderable distance frm the panel in rder t define the ang-ie frm which radiatin was being- received. A distance f 30 ft (9 m) was chsen, g-iving- an ang-ular discriminatin f 4. Fig-, 12 shws the ang-ular distributin f the radiatin; maxima can be seen at ang-ies f abut 57. A peak is als t be seen alng- the nrmal since the incident radiatin was nt randmly distributed, the maximum acustic pwer being- nearly at nrmal incidence.

15 12 Wave length In cid e n t --t--i----: radiatin Incident radiatin Wall supprting flexural vibratins ca) Fig. 11 (a) Sund incident at grazing angl e velcity f suntl in air: velcity f bending waves in wall (b) Sund incident at angle e velcity f flexural waves in wall: velcity f prgressin f wave frnts alng wall '" velcity in ai r' sin e cb) '0 27' Signal level,db Fig Plar diagram f radiatin frm a 3/16 in. (0'5 cm) aluminium sheet (warble tne f frequency ij kc/s). The lbe in the frward directin A is due t a higher level f incident radiatin in this directin. The lbes at B are due t the cincidence effect.

16 13 Results at three frequencies are shwn in Table The critical frequency has been calculated frm each measured angle, and may be cmpared with a value f 2"4 kc/s calculated frm the knwn cnstants f the panel and material. TABLE 4 Results f Measurements n Aluminium Sheet Frequency Measured Cincidence Angle 4"0 kc/s 57 4"8 kc/s 52 5"6 kc/s 44 Calculated Critical Frequency 2'83 kc/s 2"88 kc/s 2 75 kc/s These measurements were repeated at ne frequency using the crrelatr. Measurements were made with the speaker nrmal t the panel and the micrphne (a) n the same side f the panel, (b) a shrt distance away frm the panel n the ppsite side and (c) the same distance away as in the previus experiment. Three similar measurements were made with the sund incident at the cincidence angle fr this frequency. At the cincidence angle a S.R.I. 5 db lwer than that fr nrmal incidence was measured clse t the panel and 10 db lwer fr the distant psitin This indicated that the crrelatr wuld detect a narrw band f frequencies with a lw S.R.r. within the wide band used in measurement t prvide the path discriminatin fund necessary_ In the light f these cnsideratins, further measurements using the crrelatr were made at Kingswd Warren. Fur ctave bands f nise centred n 2'0, 2 8, 4"0, 5-6 kc/s were used at five angles f incidence. N indicatin f a cincidence effect was given, cnsistently high values f S.R.I" being fund. Once again, the nise cmprising flanking transmissin and ambient nise restricted the limit f measurable values The results are shwn in Table 5. Experimental Wall TABLE 5 Transmissin Lss (db) at Specified Angles Angle f Incidence Centre Frequency f Nrmal Octave Band 2 0 kc/s db 2"8 kc/s db 4'0 kc/s db 5"6 kc/s db The absence f any cincidence effect culd be due t a high value f internal damping in the panels r it culd imply that the critical frequency fr such a partitin lies abve the range f measurement. Sme similarity t "radiatin suppressin" will exist if the critical frequency is abve the audible range. This term is applied n the Cntinent and elsewhere t describe the principle f munting a light flexible panel n battens n the main massive wall; such a panel will have a high critical frequency. At lw frequencies the wall and panel will vibrate tgether and little imprvement f the insulatin will be fund. Abve the critical frequency

17 14 f the main wall, hwever, sund will pass thrugh at a cincidence angle with a lw insulatin but will nt be radiated freely due t the presence f the light panel. Prbable values fr the indices f the experimental wall have been deduced frm the preceding results and are shwn in Table 6 cmpared with the N.B.S. results and with the calculated plane wave nrmal incidence results fr a single wall f the same mass per unit area. Frequency Experimental Wall TABLE 6 Prbable Sund Reductin Indices Sund Reductin Indices (db) Experimental Wall N.B.S. Results Calculated fr single leaf wall 125 cls cls c/s c/s cls c/s "0 kc/s "0 kc/s "0 kc/s "6 kc/s Abve its fundamental resnance frequency any duble wall shuld have an increase f 5-10 db in S.R.I. ver the values fr a single hmgeneus wall f the same ttal weight. Since the fundamental resnance frequency fr this cnstructin is abut 25 cls, all measurements have been carried ut well abve this frequency. It is clear that at mst frequencies up t 2 kc/s an increase f the expected amunt is fund. 4. CONCLUSIONS The results belw 500 cls measured n this experimental wall agree fairly well with thse published by the Natinal Bureau f Standards. Abve this frequency the results are higher than the N.B.S. figures. At mst frequencies up t 2 kc/s an expected increase f between 5 and 10 db is fund relative t the values calculated fr a single structure f the same ttal weight. This may be explained entirely by the duble-leaf cnstructin f the wall, withut reference t cincidence phenmena. In certain circumstances the additinal expense f a structure f this type may be justified prviding that extreme care is taken t avid flanking paths. The muntings fr the wall structure must be cmpliant enugh t allw free vibratin f the faces. The lw apparent sund insulatin f the experimental partitin, judged by cnventinal measurements and subjective bservatins, is due t the existence f

18 15 several alternative paths fr the sund which present cnsiderably less attenuatin than the partitin itself, 5, ACKNOWLEDGMENTS Thanks are due t Mr. R, Parsns fr helpful suggestins during the cnstructin f the partitin at Kingswd Warren and t Mr. N,C,H Druce wh assisted with the experimental measurements. 6, REFERENCES 1, "Sund Insulatin f Wall and Flr Cnstructins" U. S. Dept, f Cmmerce, Natinal Bureau f Standards, Building Materials and Structures Reprt 144 (1955) 1'0510 2, K, W, Gff, "Applicatin f Crrelatin Techniques t sme Acustic Measurements", J,.A.,S,A,; VL 27, 1955, 1'.236, 3e A, C,. Raes, "Tentative Methd fr the Measurement f Sund Transmissin Lsses in Unfinished Buildings", J, A, S, A., Vl. 27, 1955, 1'.98. 4, L Cremer; "Raumakustik", VL 3, 1',173. 5c Schch & Feher, "MechanislTl f Sund Transmissin thrugh Single Leaf Partitins using Small Scale Mdels", Acustica, VL 2, N. 5, 1952, p.189. EGF