28
INFLUENCE OF ANESTHESIA (ETHER, CYCLOPROPANE, SODIUM EVIPAL) ON THE CIRCULATION UNDER NORMAL AND SHOCK CONDITIONS1 By H. STANLEY BENNETT," DAVID L. BASSETT,' AND HENRY K. BEECHER (From the Departments of Pharmacology and Anatomy and the Anesthesia Laboratory of the Harvard Medical School, Boston) (Received for publication June 24, 1943) In the presence of an injury, anesthesia can aggravate an already existing condition of shock, or if shock is impending, anesthesia can pre- cipitate it. In an attempt to evaluate the importance of several of the circulatory factors involved, we have studied dogs in good condition and in shock (hemorrhage). Particular atten- tion has been given to the influence of varying depths of anesthesia on heart rate, on systolic, diastolic, and mean arterial blood pressure, on central venous pressure, and on blood flow in three vascular beds, femoral, carotid, and mesen- teric. MATERIALS AND METHODS Animals. Satisfactory data were obtained from a total of 53 mongrel dogs, weighing about 14 kgm. each. Ade- quate blood flow data were obtained from the following experimental material: 16 dogs received evipal with 17 runs (i.e., change in anesthesia depth from very light to very deep) on 11 dogs in good condition, and 9 runs on 9 dogs in shock; 9 dogs received ether, with 14 runs on 6 dogs in good condition, and 14 runs on 7 dogs in shock; and 11 dogs received cyclopropane, with 24 runs on 10 dogs in good condition and 18 runs on 10 dogs in shock. Occa- sionally, because of technical difficulties, flow data in a given experiment were not satisfactory and yet the pressure data were good. Accordingly, the pressure data are based upon a larger series than the flow data: evipal pressure data are based upon experiments on 21 dogs; ether, on 21 dogs; and cyclopropane, on 11 dogs. The animals were heparinized with Connaught Labora- tories' heparin before cannulation, and a constant intra- venous drip of heparin was maintained throughout the remainder of the experiment. Anesthesia. Sodium evipal was administered intra- venously in 10 per cent solution. Ether was administered initially by open cone induction. The neck was dissected, a tracheal cannula inserted, and ether during the rest of 1 The work described in this paper was done under a contract, recommended by the Committee on Medical Research, between the Office of Scientific Research and Development and Harvard University. 2 Lieutenant, M. C., V(S), U.S.N.R. ' Fellow of the National Research Council. the experiment was administered by means of a closed system (Foregger) with carbon dioxide absorption. Dead space was maintained near normal. Cyclopropane was given initially in oxygen through a cone equipped with a rubber diaphragm which fitted snugly around the dog's snout. After induction, cyclopropane anesthesia was maintained as with ether. After the recording camera was started, the usual pro- cedure was to allow the animal to become lightly anes- thetized, as judged by the character of the respiration and the sensitivity of the corneal reflex. As a further index of depth of anesthesia, in some of the experiments, we used the reflex contraction of the semitendinosus muscle in response to electrodal stimulation of the sciatic nerve. The activity of this reflex was a useful guide to anesthesia depth in the animals in good condition, but not a dependable one in the animals in shock. Starting from a state of light anesthesia, the anesthesia was deepened. In the case of evipal, this was accomplished by additional intravenous injection, deepening the anesthesia quickly. The volatile anesthetics were employed so as to produce a condition of deep anesthesia during a period of about 10 to 15 minutes (Figures 4, 5, 7, 8). Administration was terminated when the animal was near the point of respiratory failure. The dog was then allowed to blow off the agent, after which the procedure was repeated. The standard preparation involved the dissection of the trachea, both carotid arteries, one external jugular vein (sometimes both), the superior mesenteric or portal veins, and one femoral artery. Dissection was done with electro- cautery, with meticulous hemostatis in order to reduce blood loss after heparinization. The preliminary induction of anesthesia and the dissection usually consumed about one hour and a half. After dissection, the animal was heparinized and flowmeter cannulae inserted into the femoral and carotid arteries and superior mesenteric vein. As a rule, the flowmeter cannulation required that the blood flow through the vessel in question be interrupted for a period of 1 to 3 minutes. Cannulae were also inserted in one carotid artery and in the jugular vein and connected with the pressure manometers for recording. The heater currents and counter currents (see Appendix) were then set for each of the 3 flowmeters, and the recording camera started. These procedures consumed an additional hour; thus about 2Y2 hours elapsed from the time of anesthesia induction until recording was started. The preparation of the animals was necessarily time consuming and involved considerable dissection and trauma, yet at the end of this, 181

INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

  • Upload
    vulien

  • View
    219

  • Download
    0

Embed Size (px)

Citation preview

Page 1: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

INFLUENCE OF ANESTHESIA (ETHER, CYCLOPROPANE,SODIUMEVIPAL) ON THE CIRCULATION UNDERNORMALAND

SHOCKCONDITIONS1

By H. STANLEY BENNETT," DAVID L. BASSETT,' AND HENRYK. BEECHER

(From the Departments of Pharmacology and Anatomy and the Anesthesia Laboratoryof the Harvard Medical School, Boston)

(Received for publication June 24, 1943)

In the presence of an injury, anesthesia canaggravate an already existing condition of shock,or if shock is impending, anesthesia can pre-cipitate it. In an attempt to evaluate theimportance of several of the circulatory factorsinvolved, we have studied dogs in good conditionand in shock (hemorrhage). Particular atten-tion has been given to the influence of varyingdepths of anesthesia on heart rate, on systolic,diastolic, and mean arterial blood pressure, oncentral venous pressure, and on blood flow inthree vascular beds, femoral, carotid, and mesen-teric.

MATERIALS AND METHODS

Animals. Satisfactory data were obtained from a totalof 53 mongrel dogs, weighing about 14 kgm. each. Ade-quate blood flow data were obtained from the followingexperimental material: 16 dogs received evipal with 17 runs(i.e., change in anesthesia depth from very light to verydeep) on 11 dogs in good condition, and 9 runs on 9 dogsin shock; 9 dogs received ether, with 14 runs on 6 dogs ingood condition, and 14 runs on 7 dogs in shock; and 11dogs received cyclopropane, with 24 runs on 10 dogs ingood condition and 18 runs on 10 dogs in shock. Occa-sionally, because of technical difficulties, flow data in agiven experiment were not satisfactory and yet the pressuredata were good. Accordingly, the pressure data are basedupon a larger series than the flow data: evipal pressuredata are based upon experiments on 21 dogs; ether, on 21dogs; and cyclopropane, on 11 dogs.

The animals were heparinized with Connaught Labora-tories' heparin before cannulation, and a constant intra-venous drip of heparin was maintained throughout theremainder of the experiment.

Anesthesia. Sodium evipal was administered intra-venously in 10 per cent solution. Ether was administeredinitially by open cone induction. The neck was dissected,a tracheal cannula inserted, and ether during the rest of

1 The work described in this paper was done under acontract, recommended by the Committee on MedicalResearch, between the Office of Scientific Research andDevelopment and Harvard University.

2 Lieutenant, M. C., V(S), U.S.N.R.' Fellow of the National Research Council.

the experiment was administered by means of a closedsystem (Foregger) with carbon dioxide absorption. Deadspace was maintained near normal. Cyclopropane wasgiven initially in oxygen through a cone equipped with arubber diaphragm which fitted snugly around the dog'ssnout. After induction, cyclopropane anesthesia wasmaintained as with ether.

After the recording camera was started, the usual pro-cedure was to allow the animal to become lightly anes-thetized, as judged by the character of the respiration andthe sensitivity of the corneal reflex. As a further index ofdepth of anesthesia, in some of the experiments, we usedthe reflex contraction of the semitendinosus muscle inresponse to electrodal stimulation of the sciatic nerve. Theactivity of this reflex was a useful guide to anesthesia depthin the animals in good condition, but not a dependable onein the animals in shock. Starting from a state of lightanesthesia, the anesthesia was deepened. In the case ofevipal, this was accomplished by additional intravenousinjection, deepening the anesthesia quickly. The volatileanesthetics were employed so as to produce a condition ofdeep anesthesia during a period of about 10 to 15 minutes(Figures 4, 5, 7, 8). Administration was terminated whenthe animal was near the point of respiratory failure. Thedog was then allowed to blow off the agent, after whichthe procedure was repeated.

The standard preparation involved the dissection of thetrachea, both carotid arteries, one external jugular vein(sometimes both), the superior mesenteric or portal veins,and one femoral artery. Dissection was done with electro-cautery, with meticulous hemostatis in order to reduceblood loss after heparinization. The preliminary inductionof anesthesia and the dissection usually consumed aboutone hour and a half. After dissection, the animal washeparinized and flowmeter cannulae inserted into thefemoral and carotid arteries and superior mesenteric vein.As a rule, the flowmeter cannulation required that theblood flow through the vessel in question be interruptedfor a period of 1 to 3 minutes. Cannulae were also insertedin one carotid artery and in the jugular vein and connectedwith the pressure manometers for recording. The heatercurrents and counter currents (see Appendix) were thenset for each of the 3 flowmeters, and the recording camerastarted. These procedures consumed an additional hour;thus about 2Y2 hours elapsed from the time of anesthesiainduction until recording was started. The preparationof the animals was necessarily time consuming and involvedconsiderable dissection and trauma, yet at the end of this,

181

Page 2: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

H. STANLEY BENNETT, DAVID L. BASSETT, AND HENRYK. BEECHER

our subjects were as a rule in excellent condition with amean arterial pressure of 120 mm. Hg or better when therecording began, and characteristically remained so untilroutine bleeding was started.

Pressures. Arterial pressure was recorded from thecannulated carotid artery. Central venous pressure wasmeasured through a flexible plastic tube inserted in theright external jugular vein and passed down so that itsopen tip lay near the entrance of the superior vena cavainto the right atrium. Zero pressure base-lines for arterialand venous pressures were taken from the same level byopening the chest of the dog after death, severing thethoracic aorta and vena cava, filling the chest with waterup to the level of the entrance of the venae cavae into theatrium, and balancing the mercury columns against thislevel, a modification of a method recommended by Green(1).

Both arterial and venous pressures were recorded op-tically by means of glass membrane manometers similar tothose described by Green (2). The natural period fre-quency of the arterial systems (approximately 150 persecond) was such as to give a satisfactorily accurate record-ing of peak systolic and diastolic pressures. A plastictubing, "Saran" (Dow Chemical Company), was used forthe hydraulic connections between manometer and can-nula. This tubing was essentially non-distensible, suf-ficiently flexible, and did not crystallize and break afterrepeated bending.

In order to record at will either full pulse pressures orintegrated mean pressures, a hydraulic damping systemwas installed (Figure 1). The single lead from stopcock Awas connected to the cannula by Saran tubing. Whenfull pulse pressures were to be recorded, stopcock A wasturned to transmit the pressure wave through the upperof the 2 parallel arms directly to the glass manometer E,and stopcock B was turned off. To obtain records ofdamped pressure, stopcock A was turned to transmit thepressure wave through the lower of the 2 arms, in whicha plug of glass wool (F) had been inserted. At the sametime, stopcock B was opened in order to connect an airbellabove the stopcock to the system. Stopcock D was keptclosed at all times, except when the amount of air in theairbell was to be adjusted for purposes of varying the timecharacteristics of the damped record. Stopcock C wasused solely to connect the manometer to a pressure bottleand mercury manometer for purposes of calibration. Theentire system was kept filled with fluid except for the airbubble which was confined above stopcock B. Thesedamped pressures did not, of course, represent an averageof systolic and diastolic pressures, but rather an integratedmean equivalent to half the area under the pressure curveduring a time constant which could be varied with the sizeof the airbell. The advantages of this apparatus over awell-damped mercury manometer lie in ready adjustabilityof the degree of damping, and easy adaptability to therecording of full pulse wave forms or mean pressures fromthe same manometer. It works equally well for arterialor central venous pressure. Typical tracings of fullpressure and integrated mean (damped) pressures, withconversions from one to the other, are shown in Figure 3.

D

B

E

C'

FIG. 1. DIAGRAM OF HYDRAULIc SYSTEM FOR OB-TAINING EITHER FULL PULSE PRESSURETRAcINGS ORINTEGRATEDMEANPRESSURES

For explanation see text.

Heart rates and pressures were counted or measured fromthe optical records. Weagree with Werle, Cosby, andWiggers (3) that because of arrhythmias and variationsof systolic and diastolic pressures with individual heartbeats, no absolute values for these figures can be taken.In order to minimize the arrhythmic effects of respiration,heart rates were counted for periods which included sev-eral respiratory cycles, 10 seconds usually being sufficient.Mean systolic and diastolic figures were estimated fromthe numerous peaks occurring in a similar period. Meanarterial and venous pressures were measured accurately.,

Blood flow. This was recorded optically by means ofthe heated thermocouple flowmeter of Bennett, Sweet, andBassett, described in an appendix to this paper. Theinstrument was improved and refinements added from timeto time during the course of the study. Hence, flowrecords taken early in the study (evipal) are not as reliablequantitatively as those obtained with ether and cyclo-propane; but directional changes in the case of evipal arevalid as presented. In interpreting the data, due attentionwas paid to the limitations and inaccuracies of the flowrecording method as discussed by Bennett, Sweet, andBassett. Frequent checks were made for changes inparasitic currents, and any changes found were corrected.The points at which these checks were made show as smallbreaks in the flow records (Figures 6, 9). When routinecorrections were made, a shift appeared in the restoredline after the break.

As shown below, there was considerable variation fromdog to dog in absolute flow values under comparable con-

182

Page 3: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

ANESTHESIA ON NORMALCIRCULATION AND IN SHOCK

ditions when recorded by this method. The majority ofthe values scattered closely around the mean, but occa-sional flows far above or below this figure were recorded.Technical errors may be involved in these aberrant values,but changes in their flow on deepening anesthesia showedpercentage changes closely in line with those nearer tothe mean. Hence, it appeared justifiable to include theseflow records in our series, since the results did not conflictwith those taken with absolute flow values near to themean. Possible technical errors in flow recording throwsome doubt on the validity of the absolute values for flow,as described in the appendix; but the means cited are notin great disagreement with values for comparable vesselsobtained by other methods (4 to 8).

Weshould like to preface our presentation of the flowdata with the clear statement that we realize fully thehazards of blood flow measurement. However, theresults are sufficiently consistent to give some basis forconviction as to validity, and pending the development ofbetter methods for recording blood flow, it appears to bejustifiable to present the results tentatively, with the hopethat they can be checked subsequently when blood flowcan be measured with more accuracy.

Impedance calculations. Impedance calculations weremade to determine whether or not given flow changes werethose to be expected from changes in pressure. An in-crease in impedance ordinarily indicates vasoconstriction,while a decrease in impedance would denote vasodilatation.The impedance figures are subject to the same errors asthose for flow.

From the measurements of flow in a given vascular bedand the difference between the mean arterial and venouspressures, the hydraulic impedance of the bed was calcu-lated by the formula,

Mean Arterial Pressure (mm. Hg)-Mean Central Venous

Impedance = Mean Pressure (mm. Hg) KMean Flow in cc. per second

This calculation is derived from Poiseuille's equationand is similar to the formula commonly used for calculating"peripheral resistance," but differs from the usual formulain that central venous pressure is taken into account, andthe word "impedance" is used instead of resistance. Con-sideration of venous pressure yields a figure of increasedaccuracy, since, as Poiseuille showed, it is actually the"difference" in hydraulic pressure which provides themoving force to fluid, and in the case of the peripheralcirculation, this difference is the difference between themean central arterial and mean central venous pressures.Both pressures must necessarily be expressed in the sameunits and with reference to the same level. In dogs, thevenous pressure factor makes for only a small error whenthe animals are in good condition, but in shock, when themean arterial pressure is low, the error arising from neglectof the venous pressure factor may be as high as 10 to 15per cent.

The term "impedance" seemed preferable to "resistance"because the animal circulation, with its fluctuatingpressures and elastic vessels, bears a close analogy to an

A.C. circuit, where inductances and capacitances mayproduce a lag between voltage and current fluctuations.The term "resistance" which implies a certain special rela-tionship between pressure and flow, properly applies onlyto hydraulic systems with perfectly rigid walls or withsteady pressure head, or both.

Shock production. After one or more "control" runsfrom light anesthesia to deep and back again, with theanimal in good condition, the animal was bled from anartery or from a vein, small amounts at a time, dependingupon the condition of the animal. Initially, this usuallyvaried from one-half to one per cent of the body weightwith subsequent bleeding at 20 to 30-minute intervals inone-quarter per cent of body weight quantity, dependingupon the animal's condition. Bleeding was varied sothat over a period of an hour or more the blood pressurewould reach a shock level, which was arbitrarily definedas a mean arterial pressure of 70 mmHg, or below. Therewas always oozing from cut surfaces after heparinization,and at times, the blood lost in this way was considerable,amounting to over 100 cc. during an experiment of severalhours. Blood loss through oozing was as far as possiblecollected and measured, and included in the figures fortotal blood loss. Undoubtedly, the dissection and oozingcontributed to the shock induced by bleeding. Hence,the shock here is not pure hemorrhagic shock, but is com-plicated by some trauma from dissection and the necessaryintestinal exposure and manipulation and in part by burnfrom the cautery. When the animal had been put intoshock the anesthesia was deepened through one or morecycles as before. This was continued until the animaldied.

Records of pressures and of flows were taken continu-ously from the cannulated vessels. The flow recordingbeams were checked about every 3 minutes for drift dueto variables mentioned before, and any drift found wascorrected. The arterial pressure recording system wasshifted from damped to undamped at frequent (about3-minute) intervals during the experiment, allowing us tofollow closely changes in systolic, diastolic, and meanpressures. The central venous pressure was usuallyrecorded as damped.

RESULTS AND DISCUSSION

The tables below summarize the data on whichthe curves (Figures 2, 4, 5, 7, 8) were based.Flow and impedance factors charted and includedare expressed in terms of percentage of flow orimpedance under conditions of light anesthesia.This afforded the best means of comparing thesefactors from animal to animal. The averageswere computed by taking the flows and im-pedances during each run at each point in termsof percentage of control flow or impedance valueunder light anesthesia, and averaging the per-centages so obtained. Standard errors of themeans are included. Pressure (mm. Hg) and

183

Page 4: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

H. STANLEY BENNETT, DAVID L. BASSETT, AND HENRYK. BEECHER

heart rate data were averaged in terms of ab-solute values. The points along the time scalesduring deepening of anesthesia were takenaccording to the fraction of time elapsed betweenthe point of administering the anesthetic and thetime of its termination. During recovery,readings were taken on a straight time basis.

The figures for absolute flow showed consider-able variation from dog to dog under comparableconditions, perhaps due in part to variations inbody size of dogs and in variations of relative sizeof head or other members with the various breedsand mixtures that were used. We have cor-

rected our flows and impedances for body surface,believing that individual anatomical differenceswould be balanced amongst the several dogs inour series.

Flow in cc. per second in femoral and carotidarteries and mesenteric veins was corrected bydividing the flow by body surface in squaremeters and the corresponding impedances cor-rected by multiplying by body surface. Bodysurface was calculated from the body weight bythe Meeh formula.

Body Surface in Meters2 = 0.112X Kilograms Body Weight2/3

Table VII shows comparisons of flows in cc.

per second per sq. meter of body surface andimpedances X sq. meters of body surface, theimpedances calculated as in the formula shownabove, with the constant taken as 1. Thefigures are averages of corrected flows andimpedances from all our dogs under ether or

cyclopropane, the values corresponding to those

in the control period of light anesthesia beforedeepening (Figures 4, 5, 7, 8; Tables III, IV, V,VI).

It will be observed that the mean flow andimpedance values shown in Table VII have largestandard errors. These large standard errors

raise the possibility that technical errors in flowrecording may be operative in some cases. Asexplained by Bennett, Sweet, and Bassett (inthe Appendix), opportunity for technical errorin the flowmeter used is considerable. Thevalues referred to indicate that this flowmeterhas detected no significant differences betweenfemoral, carotid, and mesenteric flows and im-pedances in dogs in good condition under lightcyclopropane anesthesia, as compared with flowsin the same vessels under light ether anesthesiawhen one small group of animals is comparedwith another group. A comparison of the twoagents in a given animal might still show charac-teristic differences. It would be desirable torepeat the measurements when more accuratemethods of recording blood flow become avail-able, as it would also be desirable to study theeffects of two agents in the same animal. Thepercentage changes of blood flow and impedanceare more consistent and probably more usefulthan the absolute values shown.

Sodium evipalEvipal was the first of the agents studied in

the course of these experiments, when the flowrecording technic was less accurate quantitativelythan later. However, directional changes inflow after evipal injection were consistent, and

TABLE I

Sodium evipai-dog in good conditionMeans with standard errors

Arterial blood pressure Flows in percentages of control level

Minutes VnoV us Sue

-

S r Heartafter Vrenous Sueir rteinJectionSstolic DiatpressurenFemnoral Carotid mesentericlflJSctiflSystoicDiastlic Meanartery artery or portalvein

mm. Hg mm. Hg per mixuteControl-before injection 133i:5 9645 J 10544 -3.340.7 100 100 100 171414

Sodium evipal-10 per cent solution 1.440.1 cc. i.v.

Maximum effect 1.040.2 9049 5246 69410 -2.040.8 43 :13 53411 6146 183418Recovery 4.040.3 105+8 7746 8742 -2.6E0.6 50=015 701-10 78-9 184413

184

Page 5: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

ANESTHESIA ON NORMALCIRCULATION AND IN SHOCK

.-.FEMORALx- -MESENTERIC&-CAROTID

FLOWS %CHANGES

ARTERIAL BLOOD PRESSURE

ENRAL VENOUJS PRESSUREMKHGI I I I

0 1 2 3 4TIME- MINUTES

GOODCONDITION _EVIPALFIG. 2. ACUTE EFFECTS OF EvIPAL INJECTION ON FLOWSAND PRESSUREsIN

DoGS IN GOODCONDITION AND IN SHOCKPoints are averages of pressures in mm. Hg or of flows in percentages of control

values before injection (see Tables I and II).

since, at this time, our primary interest is inqualitative effects, we have included these data.Wefrequently checked flow changes in an arterywith simultaneous records from the correspond-ing vein (Figure 3) and since the correlation was

good, we regard our recorded changes as accuratein direction and approximately correct as topercentage, and not due to changes in pulse wave

form or artefacts which would scarcely be likelyto affect both artery and vein alike.

Variations in anesthesia depth. The results ofdeepening the anesthesia with evipal in animals,both in good condition and in shock, can be sum-marized very briefly (Tables I, II, Figure 2):reduction in systolic, diastolic, and mean arterialpressures; rise in central venous pressure;

TABLE 1I

Sodium evipal-dog in shockMeans with standard errors

Arterial blood pressure Flows in percentages of control level

Minutes Venous ~~~~~~~~~~~~~~~Heartafter VeosSuperior rtpinjection l_ Dlpressure Femoral Carotid mesenteric2IJtO1 Systolic Diastolic Mean artery artery or portalvei

mm. Hg mm. Hg per minuteControl-Wore injection 102-9 1 414-4 5844 -2.6 40.4 100 100 216110

Sodium evipal-10 per cent solution 1.240.1 cc. i.v.

Maximum effect 1.2 0.2 69-7 24=3 3144 -2.640.4 1 67410 71 5 219+19Recovery 4.0-4°0.5 82-9 3144 45-5 -0.7-0.9 80+4 8146 2264:8

100 IS580

60

40

40

20

1003so

60 1.

40 +

20

_l E-I-2-3 -

SHOCK

185

x&=

I

11

d

:jIt

04

0

1%

%-%.. '! . " - - A:-

Page 6: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

H. STANLEY BENNETT, DAVID L. BASSETT, AND HENRYK. BEECHER

H0E

I

D

A

BC

. . . . . . . . . . . . . .E . . .

EVIPAL 9. 0 C.C. I.V.FIG. 3. OPTIcAL TRACINGOF ACUTEEFFECTSOF INJECTION OF EVIPAL, 10 PER CENTSOLUTION,

1.0 CC. INTRAVENOUSLY, AT BREAK IN ARTERIAL PRESSUREBASE-LINEKey to figures in all tracings. A = Arterial blood pressure, either undamped or damped.

B = Arterial pressure base line. C = Time: small marks, 10 seconds; large marks, 1 minute.D = Central venous pressure, either damped or undamped. E = Venous pressure base-line.F = Femoral artery flow. G = Carotid artery flow. H = Superior mesenteric vein flow.I = Jugular vein flow. J = Flowmeter base-line.

diminution in flow through the carotid, femoral,and mesenteric vascular beds, as recorded eitherin the artery or in the vein; no consistent changesin heart rate. Impedance calculations were notmade with the evipal flow data, since these flowvalues were not considered dependable in aquantitative sense.

The mean arterial pressure changes found hereconfirm those described by Das (9) and others.The maximum effect on the circulation usuallyoccurred about a minute after the intravenousinjection, with all components recovering gradu-ally thereafter over a period of many minutes,oftentimes not reaching again the original controllevels.

The only important difference between theacute effects of evipal injections in animals ingood condition as compared with animals inshock was that in shock, recovery from theeffects took considerably longer than in animalsin good condition and was almost always in-complete. These effects and differences are

summarized in Tables I and II and charted inFigure 2 with time relationships shown. Atypical record of simultaneous flow and pressurechanges is shown in Figure 3.

Progressive changes with shock. As shockprogressed (Figures 11, 12) with the subjectunder light evipal anesthesia, the systolic, dia-stolic, and mean arterial pressures fell asexpected. Under evipal, a curious differencefrom the other agents was apparent: Startingfrom an initial normal pulse pressure, as shockprogressed, the pulse pressure became progress-ively larger as a result of the diastolic pressurefalling more rapidly than the systolic. Thewidest pulse pressure was found usually with amean arterial pressure of 40 to 50 mm. Hg.During the agonal stage, the pulse pressurenarrowed rather rapidly.

EtherVariations in anesthesia depth. When flow

and pressure were recorded in animals in good

186

Page 7: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

ANESTHESIA ON NORMALCIRCULATION AND IN SHOCK

160

140

120

80

100 xm

80

60

220

200

180

160

80

60

40,

20

60

40

120

-4

so

-20

- 3

.-FEMORALx- -MESENTERIC'-CAROTID

X X

K- .-1 -- _--fa

IMPEDANCE %CHANGESI

fLOW CH .S..

FOW % CHANGES 0

KR

x

VENOUSPRESSURE- MM. HG

- INDUCTION 17 * 2 MIN.-

I

,x~~~~~~~~~

v_^,' - s

X x

x

-aam"am-

ETHER:- GOODCONDITION

FIG. 4. PREssuRE, FLow, AND IMDANCECHANGESIN DoGs IN GOODCON-DITION UNDERETHERANESTHESIA, WHENCARRIEDFROMA CONTROLPOINT UNDERVERYLIGHT ANESTHESIATOA DEEPLEVEL AT ORCLOSETO RESPIRATORYFAILURE

The actual time of ether administration varied somewhat from dog to dog, andpoints have been charted along the abscissa according to the fraction of the timeelapsed between the point of starting ether and the time of termination. Abscissaeduring the recovery period are taken on a straight time basis (see Table III).

condition under light ether anesthesia, and thelevel of anesthesia then deepened by givingether until respiratory failure was approachedor achieved, the following changes were observed(Table III, Figures 4, 6A): The central venous

pressure progressively rose; the systolic, dia-stolic, and mean arterial pressures usually rose

somewhat initially and then fell; blood flow inthe femoral and carotid arteries and in thesuperior mesenteric vein tended to diminish; thehydraulic impedance in these vascular beds

increased. The reduction in flow with deepeningether anesthesia confirms a similar observationby Mann, Herrick, Essex, and Baldes (10). Therise in arterial blood pressure found early on

deepening the anesthesia probably correspondsto a similar rise described by McAllister andRoot (11).

In animals in good condition, discontinuanceof the ether supply brought about the followingchanges: The central venous pressure fell ratherrapidly towards the original level. The arterial

187

X'

Page 8: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

H. STANLEY BENNETT, DAVID L. BASSETT, AND HENRYK. BEECHER

pressure recovered rapidly, often reboundinginitially above the level at which it finally sta-bilized. Flow in femoral and carotid arteriesusually showed transient increases for a minute

or more before tapering off at a stable level; con-

versely, the mesenteric vein flow tended toshow a decrease followed by recovery (Figures 4,6B). The hydraulic impedance of femoral and

TABLE III

Ether-dog in good conditionMeans with standard errors

Arterial blood pressure Flows in percentages of control level

Venous pressureSueir HataeSystolic Diastolic Mean Femoral Carotid mesentericartery artery or portal vein

mm. Hg mm. Hg per minuteControlPoint 1 148+1212 7848 94±99 -2.640.1 100±0 100±0 100±0 200±10

Ether started

2 143±-12 76±-7 94± 10 - 2.6±-0.3 87- 13 97±44 80±t153 150±t13 83±-6 107±-8 -2.04-0.3 76±-10 91±t3 92±t5 221±-404 152±-10 86±-5 107±-6 -2.1±t0.3 77±t11 100±-8 91 ±-6 203±t25 155±-9 85±-3 110±-5 -2.0±t0.3 67±-9 93±-5 82 ±-7 176±-286 140±t8 73±t1 93±t3 -2.1-±0.1 64±t11 96±t7 734-12 191±4107 141±8 71± 1 94±6 -1.5±0.3 55±-15 78±9 77±-6 202±19

Ether stopped

8 133±-2 62±t8 83±-7 -0.74-0.4 69±-9 84--18 72±t9 197±t159 143±t8 68±-5 91±t6 -2.2±t0.7 1004-13 106±-23 64±t13 229±-7

10 138±-8 53±-8 78± 7 - 2.2 ±t0.4 85±-11 76±--17 79±-21 1874-1711 134±-7 53±-7 74--7 -2.1 ±-0.3 85±-10 65±-13 85--7 2184-1412 139±3 60±14 80±11 -2.2±0.3 82±11 68±12 111±30 228±18

TABLE IV

Ether-dog in shockMeans with standard errors

Arterial blood pressure Flows in percentages of control level

Venous pressureSueir HataeSystolic Diastolic Mean Fery artid mesentericartery artery or portal vein

mm. Hg mm. Hg per minuteControlPoint 1 100±4 37±44 56±3 -3.5±0.3 100±0 100±0 100±0 197±13

Ether started

2 99±-6 42±-4 58±-6 - 3.4 ±-0.6 90±t8 98 ±t 7 100±-3 1954- 153 99±-3 39±-4 56±-4 -3.3±40.4 90±5 97±13 96±t7 198--154 90±t4 32±t4 50±t3 -2.9±t0.4 83±t9 84±-5 77±-8 1944-205 93±5 38±6 54±4 -2.6±0.4' 86±5 86±6 79±10 212±21

Ether stopped

6 76±-4 32±t2 40±t3 -1.8±t0.5 68±-8 69±t7 67±t17 208±-327 92±t12 34±t7 50±4O7 - 2.2±t0.6 94±-13 94±-14 69±-16 210±-278 75±t8 28±-3 40±t3 -2.6--0.7 85±=9 79±t6 60--12 142±t189 82±9 26-±3 40±i2 -3.4±40.3 92±-8 82±-6 87 ±9

10 73±8 29±4 43±5 -3.2±0.4 78±9 95±8 71±7 183±39

188

Page 9: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

ANESTHESIA ON NORMALCIRCULATION AND IN SHOCK

a _-FEMORALx- -MESENTERIC- #0ROTID

I-o-W - 1

80 IIMPEDANCE X CHANGES

LOW X CHANGES

H{R x- - - -x - - - - - - - -x- - - - - -

1%K

I

K- X .----- - - -

x

K

ARTERIAL BLOODPRESSURE8a HEART

GNTRALVENOUSPRESSURE-MM. HG i

INDUCTIN 13± I MI -) 4-RECOVERY 12 MIN.

ETHER:- SHOCK

FIG. 5. PREssuRE, FLow, AND IMPEDANCECHANGESIN DoGs IN SHOCKWHENCARRIED FROMLIGHT ETHERANESTHESIATO A DEEPLEVEL BY AD-MINISTRATION OF ETHER

Layout and interpretation of chart as for Figure 4 (see Table IV). Theseries of points immediately before "ether stopped" is omitted from Table IV.

carotid vascular beds tended to decrease im-mediatelyyafter removing the ether, and thento rise againttowards original levels. Themesenterichimpedance, conversely, tended to risewhile the#others fell, and then to fall againtowards the starting value.

In shock (Table IV, Figure 5), the effects ofdeepening the anesthesia were comparable to thepreceding except that: The rise in central venous

pressure was usually not as high as previously.The initial rise in arterial pressure was usuallylacking and the subsequent fall was of greatermagnitude than before, with a considerabledecrease in pulse pressure (Figure 5).

In shock, the sequence in recovery of a lightlevel of anesthesia was almost the same as ingood condition, except that events were drawnout a little in time. The heart rate showed noconsistent or significant changes with adminis-tration of ether or recovery from it. Charts oftrends of pressure and flow are shown in Figures4 and 5, and actual records of changes withvarying depth of ether in Figure 6. Individualvariations were considerable. In some of ourruns, the initial rise in blood pressure was absent(Figure 6A). Similarly, the reciprocal flowchanges in carotid and femoral arteries on theone hand and mesenteric vein flow on the other,

1o

16

140

aoD

oo xac,

189

K)O XKa80

60

220

200 ic.

180ISO

140

120

80

60

40

20

O-1-2

Page 10: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

H. STANLEY BENNETT, DAVID L. BASSETT, AND HENRYK. BEECHER

u--rn-..-

.. .-.

. . -

-

ETHER STARTED

..- 2

D

F

A

. . . . . 0 . . 0

B

0 0 0 0 % . . .

ETHER STOPPD

FIG. 6. A. PRESSUREAND FLOWCHANGESBEFOREAND AFTER ADMINISTERINGETHER (COMPAREWITH MEANSIN TABLE III)

The first part of the record is taken while the animal is becoming lighter from a

previous deepening with ether. Trends in flows and arterial pressures are upwards,and in venous pressure, downwards. With the start of ether administration, a

downward trend in flows and arterial pressure appears and venous pressure startsto rise. In this case, the arterial pressure does not show an initial rise early in thecourse of deepening, as many of our animals do, and as shown in Figure 5.

B. CHARACTERISTIc FLOWAND PRESSURECHANGESAFTER STOPPING ETHER

After turning off the ether, a downward trend in arterial pressure is reversed, anda rising venous pressure starts to fall. Mesenteric vein flow shows evidence of a

brief episode of vasoconstriction, and the femoral artery flow, a simultaneous phaseof vasodilation, followed by a trend towards original levels. Key to tracings are

as given in Figure 3.

H..6F0

A

Ca

A.0

190

Page 11: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

ANESTHESIA ON NORMALCIRCULATION AND IN SHOCK

-INDUCTION 9*0.6 MIN.-*14--

CYCLOPROPANE:- GOODCONDITION

FIG. 7. PREssuRE, FLOW, ANDIMPEDANCECHANGESIN DoGs IN GOODCONDITIONUNDER CYCLOPROPANEANESTHESIA, WHENCARRIED FROM A CONTROLPOINTUNDERVERY LIGHT CYCLOPROPANEANESTHESIA TO A DEEP LEVEL CLOSE TORESPIRATORYFAILURE

Abscissa and general considerations are as for Figure 4 (see Table V).

found usually after removing the ether, weresometimes absent entirely, with all three flowsrecovering steadily and at about the same rate.

Episodes of deliberate deepening of etheranesthesia to a point near respiratory failureseemed to be deleterious, as flows and pressuresseldom recovered to original levels (see Figures4 and 5). Several such episodes of deepeningsometimes sufficed to put the animal into ashock-like state, even without any considerableblood loss. This was in contrast to the behavior

of the animal under cyclopropane and is inagreement with the observations of Evans andBeecher (12) who found that dogs under ethertolerated shocking procedures less well thandogs under cyclopropane.

Progressive changes with shock. With a lightlevel of ether anesthesia as shock progressed, thesystolic, diastolic, and mean arterial pressuresdeclined. The pulse pressure was initially muchwider under ether than under cyclopropane orunder evipal (Figures 11, 12), and as shock

191

Page 12: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

H. STANLEY BENNETT, DAVID L. BASSETT, AND HENRYK. BEECHER

progressed, the pulse pressure became narrower Cyclopropanethan it had been. In this case (cf. evipal), the Variation in anesthesia depth. Flow andsystolic pressure fell more rapidly than diastolic. pressure records of dogs in good condition underThis process continued until death. cyclopropane anesthesia, taken at a control level

TABLE V

Cyclopropane-dog in good conditionMeans with standard errors

Arterial blood pressure Flows in percentages of control level

Venous pressure Superior Heart rate

Systolic Diastolic Mean Femoral Carotid mesentericartery artery or portal vein

mm. Hg mm. Hg per minuteControlPoint 1 159+i5 111+5 121+6 -2.2 +0.3 | 100+0 | 100+0 1004+0 | 153+8

Cyclopropane started

2 141 -+14 97+-12 113-+-13 -1.6 -+0.5 85--8 98-+4 96--7 157-- 133 153+-5 1084-5 127-+5 -2.04--0.3 91 +-5 109-+3 102+--5 157-+74 159+-49 112+it8 131 +i9 - 1.6 -+0.5 81 --8 114+;9 112+--8 107+--145 145+-7 99-+6 120+-6 -0.7 -+0.4 91+-46 114-+6 112-+7 129+-106 147+6 103+6 123+7 -0.1 +0.4 99+13 120+8 129+7 113+9

Cyclopropane stopped

7 153 + 12 99+l10 122+10 +0.7 +0.7 80+15 119+9 123+31 93+88 153+-8 104+t7 130+i5 +1.7 --0.5 63 -+9 119--7 101+-411 89-+59 177+11 118+7 146i7 -3.0 +0.4 129+12 154+9 138i11 127+9

10 146+6 100+6 116+6 -2.8 +0.3 108+9 108+4 109+4-A 150+10

TABLE VI

Cydopropane-dog in shockMeans with standard errors

Arterial blood pressure Flows in percentages of control level

Venous pressure Superior Heart rate

Systelic Diastolic Mean Femoral Carotid mesentericartery artery or portal vein

mm. Hg mm. Hg per minuteControlPoint 1 85+2 39+2 52+2 -3.6+0.4 100+0 100+0 100+0 168+9

Cyclopropane started

2 9445 42+2 58+3 -3.6+0.4 9748 106+8 103+5 168+143 9144 4145 5543 -2.8+0.4 91+8 105+6 106+5 181+104 7944 40+3 5343 -3.4+0.4 94416 105+7 100+15 156+95 75+4 33+3 4644 -2.1+0.4 90+9 94+8 88i9 163+156 85+6 42+5 56+5 -2.8+0.5 86+111 109+10 111i+20 128+77 80+45 39+4 51+4 -2.2+0.3 75412 97+9 87+13 143+10

Cyclopropane stopped

8 6844 29+3 41+44 -2.0+0.9 72+13 72+1- 1 80+19 116+109 74+t8 38+6 51+5 -1.4+0.4 664111 84+10 78+13 122+9

10 79+5 36+3 50+3 -3.0+0.7 99+16 108+12 106+ 13 154+121 1 7744 31+3 46+3 -3.5+0.4 8749 96+7 107+18 172+17

192

Page 13: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

ANESTHESIA ON NORMALCIRCULATION AND IN SHOCK

40

+ 2 ARTERIAL BLOODPRESSUREa HEART RATE+} I 2

.CENTRAL VENOUSPRESSURE- MM. H

f INDUCTION 12 ! I MIN. * RECOVERY8 MIl.t

CYCLOPROPANE:- SHOCK

FIG. 8. PRESsuRE, FLOW, AND IMPEDANCECHANGESIN DOGSIN SHOCKWHENCARRIED FROMVERYLIGHT CYCLOPROPANEANESTHESIA TO A VERYDEEP LEVEL(SEE TABLE VI)

of very light anesthesia and then followedthrough a process of deepening with cyclopropaneto a point approaching respiratory failure, in amanner similar to that described for ether,showed the following changes (Table V, Figure7): Central venous pressure rose remarkably andprogressively, reaching levels far above thoseattained under ether or evipal. Arterial pressurewas either unaffected or showed inconstantchanges. Heart rate progressively decreased.Femoral flow tended to decrease, while flow incarotid and mesenteric beds increased. Hy-draulic impedance in the femoral bed usually

increased and, in the carotid and mesenteric,generally decreased. The lack of a constanteffect on arterial blood pressure is in agreementwith Waters (13). A slowing of the heart rateunder cyclopropane has often been reportedpreviously (see Seevers and Waters (14)). Instudying the effects of cyclopropane on thecirculation in a heart-lung preparation (dogs),Krayer and Beecher (15) observed a strikingrise in venous pressure under this agent. Vol-pitto, Woodbury, and Hamilton (16) reported arise in venous pressure in man under cyclo-propane.

193

Page 14: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

H. STANLEY BENNETT, DAVID L. BASSETT, AND HENRYK. BEECHER

* -PAa*A %Al *

OvOLO kTOP

. . . . . . . . . . . .

OVOLO STOPMD*. . .. .X.1T.

CYOU STOWE

i

0

A

a3a

300

p

A

aa

FIG. 9. PRESSUREAND FLOWCHANGESAFTER ADMINISTRATION OF CYCLOPROPANEHAD ALREADYBEEN STARTED

The 2 records show the later phases of deepening anesthesia with cyclopropane, thetime of stopping the administration, and the recovery. The upper record shows a typicaldecrease in femoral flow, increase in carotid flow, rise in venous pressure, and slowing ofheart rate on deepening anesthesia, with characteristic rebound in femoral and mesen-teric flow on recovery. The marked rise in arterial pressure with deepening is atypical,and the decrease in mesenteric flow in deep anesthesia is greater than usual. The lowerrecord shows typical rising venous pressure and constant arterial pressure on deepening,with the typical rise and fall of arterial pressure on recovery. Femoral flow changes areless marked than usual.

Upon removal of cyclopropane and on allowingthe animal to recover, a rather complex series ofchanges characteristically occurred, as follows:The elevated central venous pressure fell rapidlyto a point slightly below its original value andgradually rose to a normal level for light anes-thesia. The arterial pressure sometimes fellbriefly but usually approached the original con-dition only after a sharp rise above its final

value, described previously by Brace, Scherf,and Spire (17). Heart rate was restored rapidlyto normal. Femoral flow decreased sharply fora short time and then showed a phase of increasefrom which it diminished to a stable value.Carotid artery flow showed a less markedtransient wave of increased flow, after which itfell slowly to normal. Mesenteric vein flowusually showed changes similar to those of the

194

Page 15: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

ANESTHESIA ON NORMALCIRCULATION AND IN SHOCK

..vv

* S:O esosode. , ,*~ X .* q.* obeSe....e . 0

FIG. 10. OPTICAL TRACINGSOF A DoGUNDERETHERANESTHESuTAKENAT A TiiE WHEENFLOWIN THE FEMORALARTERYWASSHOWINGIRREGULARAND UNPR.EDICTABLE CHANGES, WITHOUTEvI-DENCEOF COMPENSATINGFLOWCHANGESIN CAROTID OR MESENTERICVASCULARBED, OR OF COR-RESPONDINGCHANGESIN CENTRALARTERIAL OR CENTRALVENOUSPRESSURE

femoral, but with smaller excursions. Hydraulicimpedances showed an increase followed by adecrease, a phenomenon most marked in thefemoral vascular bed, less marked in the mesen-teric, and questionably present in the carotid.

The rise in central venous pressure, the fall inheart rate, and the increase in carotid andmesenteric flow and the depression of femoralflow during deepening anesthesia were quiteconstant. Arterial pressure changes duringdeepening were variable; sometimes no changeappeared. More often there were unpredictablephases of increased or decreased pressure.

On recovery, the fall in venous pressure wasalways present, but the arterial pressure some-times fell steadily (Figure 9A) or rose and thenfell (Figures 7, 9B), or changed very little. Thechanges in flow early in recovery were quitevariable in degree; examples of some variationsmet are shown in Figure 9. The time relation-ship which the flow and pressure changes boreto each other on recovery were also quite vari-able. Thus, the phase of vasodilatation, charac-terized by the increase in femoral and mesentericflow, occurred in some cases while the pressurewas falling, as in Figure 9A, or partly during a

phase of rising pressure and partly during aphase of fall, as in 9B, or entirely during a rise.Similarly, flow changes in one vascular bed werenot necessarily accompanied by parallel changesin another, as in Figure 9B, where the carotidflow changes were entirely out of phase with themesenteric and femoral.

In shock, the changes on deepening the anes-thesia (Table VI, Figure 8) were comparable tothose found in the animal in good condition,except that: The central venous pressure startedlower and did not rise to the same high level asformerly. The arterial blood pressure wasusually somewhat depressed when the anestheticwas pushed to a deep level. The heart rate wasa little faster throughout than it had been duringgood condition. As formerly, slowing occurredwith deepening anesthesia. The flow changeswere directionally the same, but showed lessviolent swings on recovery and the phases weresomewhat prolonged.

These episodes of deepening with cyclopro-pane, in contrast to those with ether, appeared tohave no particularly lasting deleterious effects onthe dogs. Flows and pressures usually returnedto control levels upon recovery from deepening,

195

Page 16: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

H. STANLEY BENNETT, DAVID L. BASSETT, AND HENRYK. BEECHER

199 9,9 8,9 - 7,9 - 6,9100 90 80 70 60

MEAN PRESSUREFIG. 11. PULSE PRESSURECHANGESWITH SHOCKUNDEREVIPAL, ETHER, AND CYCLOPROPANEThe dots represent the averages of all readings within the mean arterial pressure range shown on the

abscissa. The crosses denote the limits between which half of our pulse pressure readings lie, and hencegive an indication of the scatter.

and in many dogs, episodes of deepening were

imposed 5 or 6 times without any noticeabledeterioration of the circulation of the animal.

Progressive changes with shock. With a lightlevel of cyclopropane anesthesia, as shock pro-

gressed, the systolic, diastolic, and mean pres-

sures declined. The pulse pressure was initiallynarrow, as under evipal, and remained fairlyconstant throughout the shock process (Figures11, 12), with the exception that occasionally ananimal showed some widening of the pulsepressure as the mean blood pressure approachedthe shock level; but in these cases, as the shockprogressed, this would narrow again to approxi-mately the original size.

GENERALCONSIDERATIONS

The optical tracings (Figure 12) show an

example of a further difference in the nature ofthe pulse wave with the 3 anesthetics. Withanimals in good condition under ether, the meanpressure is well below a point halfway betweensystolic and diastolic pressure. Under evipal

and cyclopropane, the mean pressure is nearerthe midpoint between the systolic and diastolicpeaks.

The data obtained (Table VII) give no evi-dence of different flow values under ether andcyclopropane, respectively, in these 3 vascularbeds; accordingly, one is tempted to surmise thatthe cardiac output under these 2 agents mightbe likewise comparable, and in fact, Blalock (18)and Robbins and Baxter (19) have shown re-spectively that the cardiac output is greatlyincreased both under ether and under cyclo-propane. If this speculation be valid, one can

couple it with the lower heart rate under lightcyclopropane than under light ether and concludethat the difference in pulse pressure observed is,in all likelihood, due to differences in the elasticproperties of the aorta and large arteries withthese 2 agents. This would mean that thehydraulic capacitance4 of the aorta and large

4Hydraulic capacitance refers to distensibility, in asense comparable to electrostatic capacitance, and not tovolume.

x

X. x

80

60

40

p 20U 0LE 100Ep 80R 60ES 40SU 20RE o

60

40

20

0

196

Page 17: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

ANESTHESIA ON NORMALCIRCULATION AND IN SHOCK

arteries is greater under cyclopropane thanunder ether. In other words, the elastic tensionor Hooke's law constant of the aorta should, ifthis view holds, be greater under ether thanunder cyclopropane and it would appear to serveas a more effective damping device under cyclo-propane than under ether. It is possible thatdifferences in the time of systolic upswing mightin part impair the validity of these speculations.

Rhythmic sinusoidal fluctuations in arterialblood pressure (Traube-Hering waves) werecompletely absent in all 25 of our dogs underether anesthesia. They were present in 5 of 11dogs under cyclopropane and in 14 out of 27dogs under evipal. Although present in abouthalf the dogs under evipal and cyclopropane,Traube-Hering waves were most persistent andof greater amplitude under evipal than undercyclopropane. If such waves were present atthe time of injection of evipal intravenously, theyinvariably disappeared with the other acuteeffects of the injection, often to reappear gradu-ally with recovery after 8 or 10 minutes.

Our records of mean central venous pressureshowed in most cases a tendency towards pro-gressive decline during the shock process, with aterminal rise. This fall in central venouspressure generally observed is at variance withthe findings reported by Werle, Cosby, andWiggers (3), who, however, did not measure themean central venous pressure, but took theirvalue from an arbitrary point on the venouspulse curve. The progressive venous pressuredecline could be interrupted or reversed tem-porarily by deepening the anesthesia by adminis-tration of ether, cyclopropane, or evipal. Apartfrom the varying duration of the effects ofdeepening anesthesia with various agents, thedecline of central venous pressure with shockoccurred under all 3 of these substances.

The nature of the agonal changes in animalsdying in hemorrhagic shock differed accordingto the anesthetic used. Under cyclopropane orether, the dying animals usually showed a pro-gressive decline in systolic peak pressures withslowing of the heart rate until complete circu-latory standstill was achieved. In contrast,animals dying of shock under evipal showed asimilar initial slowing of the pulse and fall inpressure, followed characteristically by a brief

burst of cardiac recovery, with restoration ofblood flow and transient fall in central venouspressure. The recovery might last for 1 or 2minutes and then fade out with resulting finalcardiac standstill, or, in some instances, regressonly to flare up again. The cardiac slowing be-fore the reactivation observed under evipal wasin some dogs not marked, but in about half (12out of 26) of our animals under evipal, the heartslowed to complete standstill for a period of from10 seconds to a minute, and then recovered asdescribed for a minute or two. Such "recoveries"were not observed under ether or under cyclo-propane.

When arterial and central venous pressureswere measured carefully from the same zerobase-line at the level of the opening of the venacava, as explained previously, even in an animaldead from hemorrhagic shock, the arterial pres-

200-

150--100-

EVIPAL

4 a * 4

-Iuw

50o J ETHER0*

-100'0

50

-o

CYCLO-PROPANE

.. eV 0 a . a *

OOD

CONDITIONSHOCK

FIG. 12. OPTICAL ARTERIAL BLOODPRESSURE, MEANANDFULL PULSE PRESSURESWITH THE ANIMAL IN GOODCONDITION AND IN SHOCK, UNDEREVIPAL, ETHER, ANDCYCLOPROPANE

Each pair shows records from the same animal, taken atdifferent times during an experiment.

197

Page 18: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

H. STANLEY BENNETT, DAVID L. BASSETT, AND HENRYK. BEECHER

TABLE VII

Tabulation of body surface of dogs, and flows and impedances, corrected for body surface, in dogs under light cyclopropaneand under light ether anesthesia, in good condition and shock

Body Femoral Carotid Mesenteric Femoral Carotid Mesentericsurface flow flow flow impedance impedance impedance

cc. per second cc. per second cc. per secondsq. m. per sq. m. per sq. m. per sq. m.

Cyclopropane-dog in good condition

0.664O0.02 0.4340.18 1.4 A0.3 4.440.6 630d150 130=15 3849

Ether-dog in good condition

0.7340.02 0.37--0.08 1.3 -0.2 j 3.5 -1.0 4604-230 120OW40 39+7

Cyclopropane--dog in shock

0.6840.02 0.2040.07 0.5140.01 1.840.3 5204-110 130425 3617

Ether--dog in shock

0.704-0.03 0.2940.08 0.66-40.08 1.840.3 4204200 96d12 47-8

sure never fell to zero, but came to rest with thevenous pressure at 2 to 5 mm. Hg above thezero level mentioned. No significant differencesin heart rate changes between the 3 anestheticagents were observed as shock progressed.

As one can readily see from the flow changesrecorded in Figures 6 and 9, femoral, carotid, andmesenteric flows may not behave alike when all3 are recorded simultaneously. Certain differ-ences in the reactions of the femoral flow fromthose of the carotid and mesenteric show likewisein the chart of the averages under cyclopropanein animals in good shape (Figure 7). It isfurther apparent from the records that there isno consistent correlation of vasoconstriction orvasodilatation in any of these 3 vascular bedswith changes in arterial blood pressure or incentral venous pressure.

Spontaneous changes of considerable magni-tude have been recorded in one or another ofthese vessels, without any evidence of com-pensatory changes in the other beds, or anyrepercussions in systemic pressure. An exampleof such changes is seen in Figure 10, wherefemoral flow shows considerable spontaneous andtotally irregular flow changes without any signof corresponding fluctuations in carotid or mesen-teric flows or in pressures. Regular phasic flowchanges of a similar nature have been recorded.

In our experience, such phasic or irregular flowchanges have been met with most frequently inthe femoral artery. The flow in the carotid andmesenteric vessels has appeared to be more stable.These effects have been observed under all 3 ofthe anesthetic agents.

Insofar as the primary anesthetic effects go,the 3 agents considered here have great clinicalvalue, notwithstanding considerable variation intheir particular fields of usefulness. Pertinentto the problem of determining the best anestheticto use in men in shock is a comparison of thesecondary effects of these agents, particularlyon the circulation. In this connectioni, severalfacts stand out; for example, our findings suggestthat cyclopropane is tolerated better by dogs inshock than is ether or evipal. This confirms theprevious observation of Evans and Beecher (12).It may in part be due to the observed fact thatthe carotid and mesenteric flows and the arterialpressure were preserved well even when thecyclopropane was pushed to a deep level ofanesthesia. This was true with cyclopropane inanimals in shock as well as in good condition;but was not the case with ether or evipal. Fromthis, one might generalize that cyclopropaneaffords a greater margin of safety with respectto blood pressure and blood flow preservationin essential vascular beds than do the other 2

198

Page 19: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

ANESTHESIA ON NORMALCIRCULATION AND IN SHOCK

agents. These findings in dogs suggest thedesirability of searching for a similar relationshipin man. The significance of the extraordinaryrise of venous pressure under cyclopropane is notclear from our experiments, and should bestudied further. In evaluating cyclopropane foruse in wounded men, one must not lose sight ofa factor not included in the present study,namely, the well-recognized tendency of cyclo-propane to sensitize the heart to stimuli whichproduce ventricular fibrillation. This accountsfor a disturbing number of clinical deaths.

As well recognized clinically and as shown inthis and other experimental studies, anesthesiamay be deleterious in shock. Since this is thecase, the briefer the period of anesthesia, thebetter it is for the patient; accordingly, therapid induction of cyclopropane anesthesia andthe swift recovery from it offer a considerableadvantage over ether, in this respect.

Under evipal, the tendency of the centralvenous pressure to rise, and the usual failure ofarterial blood pressure and flow to return tooriginal values after fairly small injections ofevipal, support the view that this agent hadbest be reserved for brief anesthesia in subjectsin good condition.

SUMMARY

The effect on the circulation of 3 representativeanesthetic agents has been studied in dogs withthe purpose of obtaining objective informationas to the best choice of anesthetic agent for usein seriously wounded men. The influence ofvarying depths of anesthesia on heart rate, onsystolic, diastolic, and mean arterial blood pres-sure, on central venous pressure, and on bloodflow in 3 vascular beds, femoral, carotid, andmesenteric, has been studied in dogs in goodcondition and in shock (hemorrhage).

Sodium evipal effects were as follows: Deepen-ing the anesthesia of subjects in good conditionand in shock was followed in both instances byreduction in systolic, diastolic, and mean arterialpressures, by rise in central venous pressure, bydiminution in flow through the carotid, femoral,and mesenteric vascular beds. There were noconsistent changes in heart rate. The maximumeffect on the circulation usually occurred abouta minute after the intravenous injection. All

components thereafter gradually approachedtheir original levels; but when shock was present,the recovery process was slower than formerlyand the original levels were not usually attained.

Under evipal, as shock progressed, diastolicblood pressure fell faster than systolic, giving apulse pressure which widened progressively untilthe terminal stages.

Ether effects were as follows: With the subjectsin good condition, deepening of the anesthesiawas associated with a progressive rise in venouspressure; a slight initial rise in systolic, diastolic,and mean arterial pressure, followed by a fall;a diminution in blood flow in the femoral andcarotid arteries and in the superior mesentericvein, with a rise in the hydraulic impedance inthese vascular beds. Discontinuance of theether supply was followed by a rather rapid fallin venous pressure towards the original level;arterial pressures, flows, and impedances re-covered rapidly.

With the subject in shock, the effects of deep-ening the anesthesia were similar to those inanimals in good condition. During recovery,events were comparable but somewhat drawnout. An important difference from cyclopropanewas evident here: Repeated deepening of theether anesthesia was plainly deleterious; ithastened the development and increased thedegree of shock, as measured by the increasinginability of the circulatory components to returnto their preceding levels. As shock progressedunder ether, the systolic blood pressure fell morerapidly than the diastolic, giving a continuouslynarrowing pulse pressure. The pulse pressurein dogs in good condition under ether was muchwider than under evipal or cyclopropane.

Cydopropane effects were as follows: With thesubjects in good condition, on deepening theanesthesia, the central venous pressure roseremarkably and progressively. Arterial pressureshowed inconstant changes. Heart rate pro-gressively decreased. Femoral flow usually de-creased, while that in the carotid and mesentericbeds increased; the hydraulic impedance changeswere the inverse of the respective flows. Re-peated deepening of anesthesia with cyclopro-pane, in contrast to ether, appeared to have noparticularly lasting deleterious effects. In gen-eral, the changes on deepening the cyclopropane

199

Page 20: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

H. S. BENNETT, W. H. SWEET, AND D. L. BASSETT

anesthesia in dogs in shock were comparable tothose observed with the animal in good condi-tion. A complete series of flow and pressurechanges on recovery from deep cyclopropaneanesthesia was observed and described. Asshock progressed under cyclopropane, systolicand diastolic blood pressure fell at about thesame rate, without a significant change in pulsepressure, until terminal stages were reached.

In so far as one can judge from these experi-mental observations on dogs in shock as well asin good condition, cyclopropane might offer awider margin of safety for anesthesia in theseriously wounded than ether or sodium evipalwith respect to blood pressure and blood flowthrough essential vascular beds.

APPENDIX-A HEATEDTHERMOCOUPLE

FLOWMETER5

H. STANLEY BENNETT, WILLLM H. SWEET,AND DAVID L. BASSETT

(From the Departments of Pharmacology and Anatomy ofthe Harvard Medical School)

This appendix describes a flowmeter for recording bloodflow in cannulated vessels on anesthetized heparimnzedanimals. The instrument records continuously and auto-matically, and is sufficiently simple in operation to allowflow to be measured in 3 separate vessels simultaneously,without necessitating the exteriorization of blood vesselsor the use of hydraulic by-passes. It has been used invessels ranging in size from the femoral of a cat to theportal vein of a dog. It could be installed in vessels eitherlarger or smaller than those mentioned. The methodappears to be worth publishing for 3 reasons: first, becauseit has proved to be practical and useful in a rather ex-tensive series of experiments, as shown above; second,because its design eliminates or reduces many of the errorspresent in the thermostromuhr of Rein (20) and of Baldesand Herrick (21) (see Gregg et al. (22), Shipley et al. (23));and third, because its development and use have broughtto light some basic limitations and sources of inaccuracyin thermal methods of blood flow recording which havenot been adequately pointed out, and which could profit-ably be added to the list discussed by Gregg (22) and byShipley (23) and coworkers.

ConstructioHeater. We have employed, in a modified form, the

hot wire principle used by A. V. Hill (24), Anrep andcoworkers (25), and Machella (26). This method dependson the fact that the temperature of a wire immersed in a

6 The development of this instrument was aided by agrant from the Milton Fund of Harvard University.

moving stream of fluid or gas and heated by a constantcurrent will depend on the rate at which heat is takenaway from it by the moving stream, which in turn will bea function of velocity flow past the wire. The principlediffers from that used by Rein (20) and by Baldes andHerrick (21), who heated a moving stream of bloodbetween the 2 junctions of a thermocouple, external to ablood vessel.

The hot wire was a piece of nichrome wire 0.0015 inch(0.0381 mm.) in diameter and about 2 mm. long, with aresistance of about 2 ohms, mounted rigidly in a cannulaso as to be bathed by the moving stream of blood. Thenichrome wire was heated by direct current drawn froma storage battery.

Therkocouple. Instead of recording the temperaturechanges in the wire by following changes in its resistancewhile it is connected as one arm of a bridge, we havesoldered to it one junction of a thermocouple, in a mannersomewhat like that used by Gibbs (27) and by Schmidt andWalker (28) and as recommended by Burton (29). Theother junction of the couple was mounted on the oppositeside of the same cannula, 1 to 4 mm. upstream, so as to bekept as close as possible to blood temperature, and so asto be unaffected by blood heated while passing close to thehot wire. The thermocouple was in series with a movingcoil galvanometer which registered the difference in tem-perature between the 2 junctions immersed in the blood.The deflections of the galvanometer were recorded optically.

The thermocouples were of wire 0.0015 inch (0.038 mm.)or 0.0031 inch (0.0794 mm.) in diameter. Three com-binations of metals were tried: iron-constantan, copper-constantan, and chromel P-constantan. Iron-constantancouples were subject to breakage from corrosion. Wewereunable to establish any clear-cut preference between theother 2 combinations. The fine thermocouple wires weresoldered to large (No. 28) copper leads, and these pointsof soldering were kept close together (but well insulatedfrom each other) and enclosed in a small silver tube so asto keep them as nearly as possible at the same temperatureand. thus avoid stray thermal currents arising from thisextraneous pair of junctions.

Joining of heater and thermocouple. The hot junction ofthe thermocouple was soldered to a point midway alongthe length of the nichrome heater in such a way as to pre-vent the thermocouple circuit from being in parallel withany part of the hot wire. In a satisfactory junction, theheating current through the nichrome could be reversedwithout appreciably changing the deflection of the gal-vanometer in series with the couple.

Fashioning of hot and cold tips. The 4 wires leading tothe hot junction (the 2 thermocouple wires and the 2portions of the nichrome heater on each side of the junc-tion) were then folded back so as to form a narrow tent orpyramid with the junction at the apex (Figure 13), andcovered with a polymerizing phenolic insulating varnish(such as Bakelite) in which powdered quartz had beensuspended in order to increase thermal conductivity. The2 wires at the cold junction were similarly fashioned into acompact tip and insulated.

200

Page 21: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

A THERMOCOUPLEFLOWMETER

A- COPP LEAD WIRES

S- STA WIRE

C- PLASTIC TUIWO

D-PLASTIC EAL

LEGEND

EC HOT JUCTION

F- COLD JUNCTON

*-.uIcmo NE

H- CONSTANTA WK

I- HOTCHROtEL OR COPPER

J- COLD CHRONELOR COPPR

K. METAL OR PLASTIC CANNULA

L- KADS

FIG. 13. DIAGRAMMATICCROSSSECTION ANDLONGITUDINAL SECTION OF FLOWMETERCANNULAWITH THERMOCOUPLETiPs MOUNTEDIN POSITION

Mounting in cannulae. The hot and cold tips connectedto the lead wires were then mounted permanently andrigidly in a metal or lucite cannula, machined so as to fitconveniently into the vessel in which flow was to bemeasured. Metal was found to be superior to lucite be-cause of its higher thermal conductivity. The tips weremounted as shown in Figure 13, each in a hole drilledthrough the wall of the cannula, with the actual thermaljunctions projecting equally about one-third of the diam-

eter into the lumen of the cannula, and with connectingwires passing around outside the cannula, as shown.. Thetips were fastened in place with a polymerizing insulatingvarnish which also provided a continuous plastic lining tothe cannula.

In order to avoid possible error from galvanic currentsarising from salt action on 2 different wires or from thepresence of a saline bridge between the hot and cold junc-tions or the wires looping around the cannula externally,all wires were well covered by a suitable non-hygroscopicpolymerzing resin.

In order to relieve the electrical lead wires of mechanicalstress, a steel stay wire was looped around the cannula andsoldered or otherwise fastened to it. The lead wires wereanchored securely to the stay wire near the cannula, andenclosed in stiff insulation for a distance of 2 or 3 cm. fromthe cannula, in order to spare the fine wires from breakagefrom repeated flexion. The stay wire was insulated and

the lead wires wound around it in a loose spiral, anchoredwith insulating varnish, and jacketed with flexible plastictubing. Leads about 100 cm. long were found convenient.All electrical wires were carefully insulated from the can-nula and stay wire and from each other, except at the hotand cold junctions.

Provision for interchangeability. The 4 lead wires fromeach cannula were soldered at their free ends to a plugwhich could be connected interchangeably with any of 3equivalent circuits. The free end of the stay wire was

fastened to the plug in such a way as to take up tensionbefore the electrical lead wires were pulled taut.

The resistance of each completed thermocouple was

measured, and a supplementary manganin resistance coil(Figure 14, Fs) made up and mounted in the plug, theresistance of this coil being sufficient in the case of eachthermocouple to bring the total resistance of the 2 elementsto the same figure for all couples. In our thermocouples,the total resistance was 28.2 ohms.

Circuit. For simultaneous recording in 3 blood vessels,3 approximately equivalent circuits were constructed as

follows:The nichrome heater (Figure 14, C) was connected

through the plug and one end of a double-pole, double-throw switch (Qi) to a low-discharge storage battery, a 1

ohm standard resistor, and a rheostat for adjusting theheater current, all in series. The current was measured

201

Page 22: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

H. S. BENNETT, W. H. SWEET, AND D. L. BASSETT

COUNTERCURRENTCIRCUITS

LEGEND

(A) NMT ,NCTO LEED SNORT STUTTMPE to LES FORwM ulRING* COLD JUbCTION sE WRE Tu O-' CMOMEMUEATERWEE 03 EERALatO OR LeSn S 001I LEWFORNEASUROIfATEtwa o z e lron ~~~~~~~~~~cuRRm,

SNORTEDIWO 4 OROUPSOF SACK "S U s S M 5 0)CWOMASURKLEAD

* W.L LOW $TOSTAInn _ '_) LEEDS* N N.40Cal (TYV W PR ems summ TYP K OOUNLE POLE TiN WNDI

LES SRTNRWpTYPE MS Los OPOT KNIE sWimONvfw44-AFOIL CC

FIG. 14. Cmcurr DuGR

and standardized by determining the voltage drop acrossthe 1 ohm standard resistor (J). Through the other endof the switch (Q), the battery could be connected inseries with a dummy heater resistance of 2 ohms (F1),through which it could be caused to discharge at the ex-pected rate for a few hours before actual use so as to allowits voltage to become stabilized.

The thermocouple leads were likewise connected througha similar switch (Qs) in series with the galvanometer (G)and certain resistors (F4 and F5). These resistors were sochosen as to make the total resistance in series with thegalvanometer equal to its critical damping resistance, assupplied by the manufacturer. The switch (Q*) containedprovision for connecting the galvanometer in series with adummy manganin resistance (F2) equal to that of thethermocouple, for test purposes.

A counter current tapped off from a balancing poten-tiometer in seies with a 2 volt low discharge storage cellwas connected as shown, in order to allow one to balance

out the relatively large basic galvanometer deflection re-sulting from the heating of the hot junction, and to bringthe beam back so that the desired fluctuations with flowwill fall on the recording scale. The voltage to be tappedfrom the potentiometer to accomplish this was determinedfor each cannula in each circuit at the time of calibrationand set at the determined figure at the beginning of eachexperiment and left constant.

In order to compensate readily for shifts in the zeropoint of the galvanometer and for variable stray parasiticthermal currents arising in the circuit or cannula, a com-pensating potentiometer was connected, as shown, througha reversing switch (Q4). With switches Qi and Qs open,and switch Q2 connected to the thermocouple, extraneousthermal EMF's which might be present and causing errorcan be detected and nullified. It has been our practice totest rather frequently for these shifts and when necessaryto use the compensating potentiometer to reset the gal-vanometer beam to its proper zero point while it is in

202

Page 23: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

A THERMOCOUPLEFLOWMETER

series with the thermocouple, unaffected by heater or

balancing current. The batteries providing power for thebalancing and for the compensating potentiometers were

kept on continuous discharge.Provision was made for conveniently and accurately

measuring the voltage tapped off from the balancing poten-tiometer, the voltage drop across the 1 ohm standard inthe heater circuit, and the actual thermocouple voltage,by means of a Leeds and Northrup Type K potentiometer,properly connected to its own battery, standard cell, andgalvanometer. This measuring potentiometer (K) couldbe readily connected to any of the leads from any of our 3circuits by means of the rotary selector switch (P).

The circuit was constructed with great care so as toconform in every possible respect with the standards neces-

sary for reliable performance in critical low-voltage D.C.circuits. All permanent connections were carefully sold-ered, all resistors were of manganin, and all switches, lugs,connecting wires, etc., were selected so as to minimize thevariations in resistance or voltages with temperature or

other variables.

Calibration

The cannulae were calibrated with defibrinated beefblood, filtered through nylon bolting cloth, run bv gravitypressure through the cannula, and measured in graduates.Calibration was carried out in a water bath at approxi-mately 38°C. Variations in bath temperature within therange of body temperature fluctuations did not affect thecalibration.

The heater current necessary to give desired sensitivityand the counter current necessary to bring the galvan-ometer beam to bear on the recording scale were deter-mined for each thermocouple in each circuit. The sensi-tivity of the couple could be adjusted by varying theheating current, high sensitivity being achieved by highheating currents. A sensitivity giving a galvanometerdeflection of 2 cm. for every doubling or halving of flowwas found convenient for most cannulae.

In calibration, a series of various blood flows were

passed through the cannula with heater and countercurrents set, and deflections of the galvanometer noted.A sigmoid curve was obtained when voltage or galvan-ometer deflection was plotted against log of flow in cc.

per second. The curves were conveniently plotted on

semi-logarithmic paper, with voltage or deflection on thelinear and flow on the logarithmic coordinates (Figure15A).

In calibration, a definite zero flow deflection, stable to

within a few millimeters, was obtained and found to berepeatable in the animal. In this respect, the instrumentdiffers from those of Baldes and Herrick (21) and of Rein(20), which have no definite zero flow deflection (Figure15D) (see Herrick, Baldes and Sedgwick (30)). Zero flowequilibrium is achieved when the heat supplied by theheating current is taken away solely by conduction throughthe cannula and fluid, and by convection. Zero flowcannot be charted on semi logarithmic paper, but the

deflection point was placed on the charts and assumed tobe the point at which the curve becomes vertical. Gal-vanometer deflection decreased progressively as flowincreased from zero.

Changes in counter current from the balancing or com-

pensating potentiometers merely moved the calibrationcurve laterally along the voltage coordinates without dis-torting the curve in any way. Figure 15A shows 2 curves

taken with the same heating current with the same cannula,but with different counter currents-the number of milli-volts tapped from the balancing potentiometer being givenfor each curve. Theoretically, the 2 curves should beentirely parallel, separated from each other by the same

distance along the abscissa for any given ordinate. Cali-bration curves departing significantly from this ideal were

discarded.The part of the curve for fast flows is very steep, and

was found to be less repeatable than other portions (Figure15B). In practice, it was usually possible by proper choiceof heating current to place this steep portion beyond therange of flows met with physiologically. The entire curve

below this steep portion could be used satisfactorily.Xery slow flows were met with only terminally in our

experiments, and we found it convenient to calibrate our

cannulae with a heating current sufficiently large to givea range of galvanometer deflections from zero to maximalflow too large to be encompassed by our recording scale,which was 12 cm. wide. Accordingly we customarilymade 2 curves with the same heating current but with dif-ferent balancing currents (as in Figure 15A). It was

possible to record almost all of the records on the high flowcurve, the low flow curve being reserved for special pur-

poses.

The calibration curves were repeatable and stable, andit was not necessary to calibrate each unit after or duringeach experiment. Figure 15B shows points from 6 calibra-tion runs with straight forward flow on a single cannula,each run being taken on separate days, scattered over

about a month's time. The sma4 degree of scatter isevident. Also charted are points taken with pulsatile flowinvolving a phase of back flow, the amount of back flowbeing given for each point in terms of percentage of netforward flow. Points taken with back flow up to 40per cent of net forward flow show no more scatter thanthe points taken with even forward flow. The reason forthis is inherent in the structure of the thermocouples andtheir placement, and is discussed below. The curves were

not affected by varying concentrations of red cells withinlimits met with in experimental animals, nor by changingthe conductivitv or the environment around the cannulain the bath, as by packing it in cotton, enclosing thecannula in a rubber tube, or leaving it free in the stirredfluid of the bath, as long as the temperature of the bathwas the same as that of the blood used for calibration.With the heater currents used in practice, the maximaltemperature of the hot junction at zero flow was 2 to 60above blood temperature, depending on the mass ofmetal and plastic close to the heater, and on the heatercurrent.

203

Page 24: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

* CANNULA 235 CCIRCUIT BHEATING CURRENT

58 MA.

30

20

lC

4

6.440 MV.

*8.360 MV..0

.0

4

.1

43

CANNULA 225 CCIRCUIT BHEATING CURRENT100 MA.

xo

0

i23% 9.000 MV..41%

ojC39%*36%

*22%0

X

* RUNS 0425%a0N oXv SEARATEX DAYS WITH

EVEN FLOWaPULSATILE FLOW

WITH % OFBACKFLOW

B

£L V

A K

x

0

*X.

* * * a I I II I I I I I I,

I . I I I I I

ZEROFLOW.C - a D. \-m

FIG. 1SA. CALIBRATION CURVESOF A TYPICAL CANNULA

Each curve was taken with the same heating current, but with different counter current-the voltage tapped fromthe balancing potentiometer being given in each case. Ordinates: Blood flow in cm.3 per second. Abscissae: Galvanom-eter deflection in cm. One mm. deflection is equivalent to about 0.5 microvolts of thermocouple voltage.

B. CALIBRATION POINTS TAKENWITH A DIFFERENT CANNULA, SHOWINGTHE AMOUNTOF SCATTEROBTAINED WITH

6 DIFFERENT RUNSWITH STRAIGHT FORWARDFLOWAND 1 RUNWITH PHASIC FLOWINVOLVING BACKFLOW

Each run was taken with the same heating and counter currents, but on different days, scattered over a month's time.Back flow is expressed in terms of percentage of net forward flow. Ordinates and abscissae as in Figure 15A.

C. RECORDOF FLOWTHROUGHA CANNULATO SHOWSPEED OF RESPONSENote that the beam comes to rest within about 1.5 seconds of the time of producing a sudden change of flow through

the cannula. Duration of time signal-3 seconds.

D. COMPARISONOF SPEED AND DIRECTIONAL RELIABILITY OF RESPONSEOF THISFLOWMETERWITH BALDES THERMOSTROMUHR

The upper beam is the record of a Baldes unit in series with one of the present cannulae which is recording on thelower beam. At the spike artefact on the lower beam a sharp decrease in flow was produced, causing an immediatedownward deflection of the beam recording from our instrument. The Baldes unit showed no change for a few seconds,and then gave a spurious deflection in the direction of increased flow, followed by a swing toward decreased flow. Thesmall spike at the beginning of downswing in the lower record is an electrical artefact, induced from the coil of the signalmagnet. Such effects can be avoided by shielding. Duration of time signal-3 seconds.

6

43

.4

2

.1

,4:I4

.0

.01

.0.o

Jo

.4

A

.

.1

.

.

.

Page 25: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

A THERMOCOUPLEFLOWMETER

Time characteristics

The mounted thermocouples had variable time charac-teristics: those with small tips mounted in metal cannulaewith little excess insulation responded more rapidly thanthermocouples with tips in lucite cannulae, or with ratherbulbous plastic coverings. At fast flows, the thermo-couples changed temperature very rapidly with changesin flow, the limiting factor in speed of response being thetime lag of the critically damped galvanometer coil, whichhad a period of 1.5 seconds (Figure 15C).

At changes to slow flows, or when flow stopped, thermalequilibrium was achieved more slowly, as under these con-

ditions an appreciable amount of fluid had to be heated.Because of the high heat capacity of water, the passage ofan appreciable number of seconds was required before theheater could provide enough calories to heat up the water

to the point where increased temperature differencebetween heater and fluid gave rise to an equilibrium in-volving loss of heat principally by conduction and convec-

tion. Thus, at slow flows or when flow stops, the limitingfactor for speed of response is the thermal lag of the fluidsurrounding the heater. The actual time required for com-

plete equilibrium to be achieved when flow was suddenlystopped varied with the nature of the cannula and insula-tion, but ranged, in general, between 5 and 10 seconds.

Cannulation

The insertion of the cannulae requires that the vesselsbe dissected, the animal heparinized, the blood vesselsclamped and cut open, the cannula inserted, filled withsalt solution, and secured, and the flow through the vesselreestablished. As a rule, the total time during which theflow through any vessel was shut off for cannulation was

from 1 to 3 minutes. This short time was made possiblelargely by the use of special ligature clamps made bydrilling the ends of mosquito snaps as shown in Figure 16.Before cannulation, 2 waxed braided silk ligatures were

passed around the vessel and threaded through the holesin the ligature clamp, as shown. After insertion of thecannula, the ligatures were drawn tight around each beadof the cannula and clamped. This required less time thanwould be required to tie knots. The flow was then rees-

tablished in the vessel, final ties were secured around thecannula, and the clamps, which thus served a temporarypurpose, were then removed.

Performance and sources of error

After cannulation, the thermocouple was plugged intothe circuit and the heater and counter currents set at thefigures determined in calibration. The precautions takenin choice of materials and batteries and in construction ofthe circuit were sufficient to give adequate stability ofheater and counter currents, provided the heater batterywas allowed to discharge at the expected rate for a fewhours before actual use. Heater current and countercurrent voltage were checked from time to time duringeach experiment, and adjustments were rarely necessary.

However, the instrument was not entirely stable and

Vo

FIG. 16. THREADEDLIGATURE CLAMPMADEBy DRILLINGMIOSQUITO SNAP

free from drift when used in animals. The bulk of theinstability was traceable to the cannula and thermocoupleitself. The feature of mounting the heater and thermo-couple tips well inside the blood stream was adopted inorder to free these elements as much as possible from tem-perature variations in the media surrounding the vessel.This feature has been only partially successful; apparently,there is some slight thermal conduction through the can-

nula wall and along the wires and insulation to the thermo-couple junctions, for the deflection of the galvanometercould be influenced somewhat by changes in the tem-perature of the environment of the cannula. The influenceof these variables was reduced by packing cotton aroundthe cannula where it rested in the animal, by closing thewound snugly over it, and by covering the general area

with cotton or blankets so as to keep the temperature as

constant as possible.Even these precautions did not suffice to eliminate drift

entirely, and it was necessary to check for, and correctelectrically, currents arising from varying ambient tem-perature by use of the compensating potentiometer andswitches Qi and Q3, as described under "Circuit." Bycorrecting for these factors as frequently as the nature ofthe experiment demanded, it was possible to get flowrecords in which the amount of drift was known or cor-

rected, with consequent improvement in accuracy. Eachepisode of checking involved disconnecting the thermo-couple from heat and counter current, with a consequentswing of the galvanometer beam to a resting place near itszero point off the recording scale. After checking and, ifnecessary, resetting the zero point, closure of switches Qland Q3 brought the beam back to its proper recordingposition, leaving a short gap in the flow recording linecorresponding to the time the galvanometer beam was offthe scale. If resetting of the compensating potentiometerwas necessary, a shift appeared in the projection of thereconstituted flow meter record. The degree of instabilityand drift ordinarily encountered, as revealed by this shift,can be seen in the flow records (Figures 6, 9). Quitepossibly, error from varying environmental temperaturemight be further reduced by refinements in the placementand construction of the thermocouple tips. However, we

feel other sources of inaccuracy are so large as not tojustify any great experimentation along such lines.

205

Page 26: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

H. S. BENNETT, W. H. SWEET, AND D. L. BASSETT

A further source of error arose from the fact that allportions of a moving blood stream may not be at exactlythe same temperature. As shown by Franklin andMcLachlin (31), blood from various tributary streamsemptying into a larger vein may remain as separate anddistinct "stream lines" for a considerable distance downthe larger vein. Such a stream from a tributary veindraining surface tissues may be at a temperature differentfrom a similar stream from deep tissues, and these 2streams may varyingly influence the 2 junctions of thecouple. Changes in the relative proportions of these com-ponents might produce galvanometer deflections whichwould not represent true flow changes.

This factor was troublesome at times in flow measure-ments in peripheral veins, such as the jugular and femoral.It has not been so troublesome in deep veins, such as theportal or mesenteric, or in arteries. It is not always pos-sible to correct for this error by means of the compensatingpotentiometer, as inflow of cold blood from a surface vein,for example, may vary phasically with respiration, or maybe changed sufficiently by pharmacological agents or otherfactors to make interpretation of the records extremelydifficult.

Error can arise within the cannula if it be angulatedwith respect to the blood vessel leading into it, so that therapid portion of the blood stream deflects against the hotjunction, giving rise to an apparent flow recording whichexceeds the true one. Conversely, if the stream is de-flected against the wall of the cannula opposite the hotjunction, the hot junction may find itself in a componentof the stream moving more slowly than normally andrecord a falsely low flow. In practice, it has usually beenpossible to align the cannula so that this deflection errorwas small or absent. At times, however, especially indeep cavities, such good alignment has been difficult tomaintain, and erroneous records have resulted. Errorfrom angulation may be considerable in short, wide can-nulae; it is negligible in cannulae which are long in relationto their diameter.

A further source of error arises from the fact that therate at which heat is taken away from a heated wire by amoving stream of fluid does not bear a linear relation tovelocity, whereas the EMFfrom a thermocouple and theresistance changes in a wire are very nearly linear functionsof temperature changes within the limits of the applicationto blood flow measurements. Hence, the mean deflectionobtained by a mechanical integration, by the galvanometer,of the thermal currents varying with phasic flow changeswill not represent true mean flows over the correspondingperiod of time. The size of this deviation would varywith the nature and period of the phasic changes, but thetest runs with the phasic flow in the calibration bath(Figure 15B) suggest that the error may be more oftheoretical than of practical importance, as far as thisinstrument is concerned.

Gregg et al. (22) and Shipley et al. (23) pointed out thatback flow during certain phases of diastole could causegreat error in the Baldes thermostromuhr. Back flow inour flowmeter was of less consequence as a source of error

than in the Baldes instrument since (1) the cold junctionwas mounted in a position where it was not affected bywarmed blood washed back from the heater, and (2) heatwas applied directly to the hot junction so that bloodwashed back on it still retained, initially at least, someheat from its recent forward passage past the couple, andhence cooled it less in back flow. As stated previously,phasic back flow up to 40 per cent of the net forward flowwas tolerated by our instrument without appreciable error.

General appraisal and comparison with other thermnal methods

This instrument was sufficiently stable to give calibrationpoints with a scatter of about i 10 per cent or i 15 per centabove or below the mean flow in favorable portions of thecurve (Figure 15B). This was better than we were ableto achieve using comparable precautions with the Baldesapparatus (cf. Gregg et al. (22)).

However, in the animal, the instrument was less reliable,and the experimental data now available do not allow usto set any rigid limits as to its accuracy. Obviously, inthe presence of angulation, or of marked difference intemperature between various components of the bloodstream flowing through the cannula, or in the presence offibrin clots, it may be totally inaccurate.

Under reasonably satisfactory operating conditions,when external temperature was not fluctuating too rapidly,when angulation was absent, and when the blood withinthe vessel was at even temperature, fairly consistentresults were obtained, since (1) flow values in a givenvessel, corrected for body surface of the animal, comparedfavorably with flows similarly taken and corrected fromother animals under comparable conditions, and (2) flowmeasured in an artery was close to flow in the correspondingartery on the other side, or with flow in the correspondingvein. Good checks between artery and correspondingvein would appear to render unlikely any error from phasicchanges or from back flow, which would scarcely be ex-pected to affect both artery and vein alike or to the samedegree. Examples of mean values of flows (with standarderrors) in various vessels, obtained during a considerableseries of experiments, are given in Table VII. Thesemean values are in fair agreement with correspondingfigures obtained by other workers with other methods,but the fact that occasional flows were recorded wellabove or below the mean suggests that technical errorsmay at times be considerable. Similar discrepancies incomparing flows in corresponding vessels in the sameanimal were also sometimes encountered.

Our instrument utilized refinements in circuit construc-tion beyond those used by Rein (20), Baldes and Herrick(21), Gibbs (27), and Schmidt and Walker (28). Theserefinements have been effective in improving the timecharacteristics, stability, and accuracy of the instrument.Attaching the thermocouple directly to the heater, as doneby Gibbs (27) and Schmidt and Walker (28), gave a stablezero flow deflection, which was lacking from the instru-ments of Baldes and Rein, as pointed out by Herrick,Baldes, and Sedgwick (30). The influence of environ-mental changes in temperature, which contribute most

206

Page 27: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

A THERMOCOUPLEFLOWMETER

markedly to the inaccuracy of the Baldes thermostromuhr(Gregg et al. (22); Shipley et al. (23)) and which wouldlikewise affect the instrument of Rein (20) and of Schmidtand Walker (28) was greatly reduced in our instrument bymounting the thermocouples directly inside the bloodstream. In addition, provision was made for nullifyingerror from this source when it was present. Errors in theGibbs (27) method, arising (1) from changes in the caliberof the vessel, and (2) from possible shift of the needle tipfrom rapid axial portions of the blood stream to slowercircumferential portions, were eliminated in the presentinstrument by mounting the thermocouples in a rigidcannula in a fixed and permanent position relative to themoving blood stream (in the absence of angulation). Byreducing the amount of inert material around the heaterand thermocouple, the time characteristics have beenimproved so that the limiting factor as to speed of responseat fast flow changes is the mechanical inertia of the gal-vanometer, and at slow flow, the thermal lag of the sur-rounding fluid.

This considerable lag of response with changes to veryslow flows or to stopped flows-the lag being a consequenceof the high heat capacity of water-appears to be a hithertounappreciated limitation of thermal flow recording influids. Machella (26) apparently allowed himself to beled into error from failure to consider this factor. He useda small heated nickel wire which would change temperaturerapidly, and which would follow fast flow changes quiteclosely. His wire operated at about 60 above blood tem-perature. When flow stopped, it was necessary for thiswire to heat the surrounding blood to nearly this tem-perature before equilibrium could be achieved, and thiswould require an interval longer than the diastolic period.Hence, the instrument was inadequate for detectingstoppage of flow in the aorta in diastole, and his denial ofthis stoppage is not warranted. Shipley, Gregg, andSchroeder (8), who obtained evidence of diastolic stoppageor even back flow in the aorta using another method, wereunable to explain the discrepancy between their results andthose of Machella. The factors mentioned here wouldprovide a possible explanation for this discrepancy.

Basic limitation of thermal methodsBecause of the numerous attempts to develop instru-

ments to record blood flow by thermal methods, withresults never wholly satisfactory, it would seem advisableto list factors which limit the usefulness and accuracy ofall thermal methods. Some of these factors have beenpointed out before with respect to the particular instrumentof Baldes and Herrick, by Gregg et al. (22) and Shipleyet al. (23).

1. Thermal methods cannot be relied upon to recordrapid phasic changes in flow in fluids involving phases ofvery slow flow or of complete stoppage. The thermal lagof fluid surrounding the heated element is such thatseveral seconds may be required for equilibrium to beattained after flow stops.

2. Uneven temperature in various portions of a givenblood stream in a vessel may be present to a degree suffi-

cient to cause temperature changes in the heated element,falsely representing flow changes of considerable magnitude.

3. The temperature of the heated element does not beara linear relationship to flow. Hence, when pulsatile flowchanges are present, an integration of voltage or resistancechanges by a recording system will not represent a truemean flow.

4. Fluid moving in either direction cools a heated ele-ment equally well, and hence, phasic back flow will recordas forward flow.

These limitations, when combined with the great diffi-culty encountered in attempting to free completely theheated element from varying environmental factors, appearto be sufficient to render unlikely the development of avery satisfactory blood flow recording system utilizing thethermal principle.

Summary

A heated thermocouple flowmeter suitable for providingautomatic and continuous recording of blood flow in can-nulated vessels of anesthetized heparinized animals isdescribed. The instrument is designed to eliminate orminimize certain factors producing error in previousthermal methods of flow recording. The advantages andlimitations of the instrument are discussed and its per-formance is compared with that of other thermal flow-meters.

Certain basic limitations and sources of error inherentin all thermal methods of flow recording appear to besufficient to render unlikely the development of a verysatisfactory blood flow recording system utilizing thethermal principle.

BIBLIOGRAPHY

1. Green, H. D., Zero blood pressure level. Am. J.Physiol., 1941, 133, P 300.

2. Green, H. D., A glass membrane optical manometer.Federation Proceedings, 1942, 1, 32.

3. Werle, J. M., Cosby, R. S., and Wiggers, C. J., Ob-servations on hemorrhagic hypotension and hemor-rhagic shock. Am. J. Physiol., 1942, 136, 401.

4. Burton-Opitz, R., The vascularity of the liver. XI.The motor reaction in the portal radicles of theliver. Quart. J. Exper. Physiol., 1913, 7, 57.

5. Rein, H., 1iber die physiologischen Aufgaben desAdrenalins als Kreislaufhormon. Verhandl. d.deutsch. Gesellsch. f. Kreislaufforsch., 1937, 10, 27.

6. Katz, L. N., and Kolin, A., The flow of blood in thecarotid artery of the dog under various circum-stances as determined with the electromagneticflowmeter. Am. J. Physiol., 1938, 122, 788.

7. Grindlay, J. H., Herrick, J. F., and Mann, F. C.,Measurement of blood flow of the liver. Am. J.Physiol., 1941, 132, 489.

8. Shipley, R. E. Gregg, D. E., and Shroeder, E. F., Anexperimental study of flow patterns in variousperipheral arteries. Am. J. Physiol., 1943, 138,718.

207

Page 28: INFLUENCE OF CYCLOPROPANE, SODIUM EVIPAL) SHOCK · cautery, with meticulous hemostatis in order to reduce bloodlossafterheparinization. Thepreliminaryinduction of anesthesia and the

H. S. BENNETT, W. H. SWEET, AND D. L. BASSETT

9. Das, S. C., The influence of sodium evipan on theheart and circulation. Quart. J. Exper. Physiol.,1941, 31, 103.

10. Mann, F. C., Essex, H. E., Herrick, J. F., and Baldes,E. J., The flow of blood in relation to anesthesiaand operation. West. J. Surg., 1935, 43, 177.

11. McAllister, F. F., and Root, W. S., The circulatoryresponses of normal and sympathectomized dogs toether anesthesia. Am. J. Physiol., 1941, 133, 70.

12. Evans, E. I., and Beecher, H. K., Unpublished obser-vations, 1940.

13. Waters, R. M., Present status of cyclopropane. Brit.M. J., 1936, 2, 1013.

14. Seevers, M. H., and Waters, R. M., Pharmacology ofthe anesthetic gases. Physiol. Rev., 1938, 18, 447.

15. Krayer, O., and Beecher, H. K., Unpublished observa-tions, 1938.

16. Volpitto, P. P., Woodbury, R. A., and Hamilton,W. F., Direct arterial and venous pressure measure-ments in man as affected by anesthesia, operation,and shock. Am. J. Physiol., 1939-1940, 128, 238.

17. Brace, D. E., Scherf, D., and Spire, L. J., The effectof cyclopropane on the blood pressure, strokevolume and heart size of the dog. Anesthesiology,1941, 2, 261.

18. Blalock, A., Cardiac output in the dog during etheranesthesia. I. The effect of ether anesthesia onthe cardiac output. Arch. Surg., 1927, 14, 732

19. Robbins, B. H., and Baxter, J. H., Jr., Studies ofcydopropane. IV. Cardiac output in dogs undercyclopropane anesthesia. J. Pharmacol. and Exper.Therap., 1938, 62, 179.

20. Rein, H., Die Thermo-Stromuhr. Ein Verfahren zurfortlaufenden Messung der mittleren absoluten

Durchflussmengen in uner6ffneten Gefassen in situ.Ztschr. f. Biol., 1928, 87, 394.

21. Baldes, E. J., and Herrick, J. F., A thermostromuhrwith direct current heater. Proc. Soc. Exper. Biol.and Med., 1937, 37, 432.

22. Gregg, D. E., el al., Observations on the accuracy ofthe thermostromuhr. Am. J. Physiol., 1942, 136,250.

23. Shipley, R. E., Gregg, D. E., and Wearn, J. T.,Operative mechanism of some errors in the applica-tion of the thermostromuhr method to the measure-ment of blood flow. Am. J. Physiol., 1942, 136,263.

24. Hill, A. V., Electrical recording of the pulse. Lancet,1920, 2, 752.

25. Anrep, G. V., Cruickshank, E. W. H., Downing, A. C.,and Subba Rau, A., The coronary circulation inrelation to the cardiac cycle. Heart, 1927, 14, 111.

26. Machella, T. E., The velocity of blood flow in arteriesin animals. Am. J. Physiol., 1936, 115, 632.

27. Gibbs, F. A., A thermoelectric blood flow recorder inthe form of a needle. Proc. Soc. Exper. Biol. andMed., 1933, 31, 141.

28. Schmidt, C. F., and Walker, A. M., A thermostromuhroperating on storage battery current. Proc. Soc.Exper. Biol. and Med., 1935, 33, 346.

29. Burton, A. C., Appendix: theory and design of thethermostromuhr. J. Appl. Physics, 1938, 9, 127.

30. Herrick, J. F., Baldes, E. J., and Sedgwick, F. P.,Experimental analysis of Rein's thermostromuhrfor small flows. J. Appl. Physics, 1938, 9, 124.

31. Franklin, K. J., and McLachlin, A. D., Streamlinesin the abdominal vena cava. J. Physiol., 1936, 86,386.

208