A New Tonometric Method for the Determination of Dissolved Oxygen and Carbon Dioxide in Small Samples

Four types of method are available for the determination of dissolved oxygen in biological studies, viz. chemical, polarographic, gasometric and tonometric. The last two types are also suitable for carbon dioxide determinations.

Micro-Winkler methods are satisfactory for samples of 1-2 ml. or larger, but accuracy is limited by the sensitivity of the titration end-point if smaller samples are used. Furthermore, errors arise if the samples contain certain dissolved substances such as nitrites or ferrous iron, which may occur as impurities in water samples. The latter limitation of Winkler’s method has been discussed by Alsterberg (1926) and by Allee & Oesting (1934), who have proposed rather elaborate modifications to deal with specific impurities. These modifications, if found to be necessary in particular cases, considerably detract from the utility of the original method.

Polarographic methods involving carefully controlled electrolysis with the reduction of oxygen have proved useful in recent years. However, constancy of pH and of the electrolyte composition are essential and even then frequent calibration against an independent standard is necessary. The paper of Brown & Comroe (1950) and the monograph of Koitoff & Lingane (1952) should be consulted for details and references.

A number of gasometric methods are available, in which a large and definable fraction of the dissolved gases is extracted from the sample and determined volu-metrically or manometrically. In the van Slyke method (van Slyke & Neill, 1924, and Oesting, 1934) the dissolved gases are extracted under vacuum and samples of about 5 ml. are required to give a reasonable standard of accuracy. Both oxygen and carbon dioxide can be determined with one sample. An alternative method (Scholander, van Dam, Claff & Kanwisher, 1955) involves extraction of the gases from the sample by mixing acid and carbonate in the sample syringe in order to generate a relatively large gas phase. This precludes the determination of carbon dioxide in the sample, but the oxygen determination can be carried out with a sample of only i ml. volume.

No chemical or gasometric method is applicable to the determination of the tensions^* of dissolved oxygen or carbon dioxide when respiratory pigments or buffer systems are present, since these methods will yield values for the total quantity of the respective gases.

Tonometric methods offer the most satisfactory approach to the determination of both oxygen and carbon dioxide in small volumes of biological fluids. Here the principle is simply one of equilibrating a relatively small gas bubble with the sample and subsequently analysing the bubble. Particular efforts have been made by a number of workers to evolve a satisfactory technique based on this principle, for the determination of the oxygen tension of arterial blood in man (see Riley, Campbell, & Shephard (1957) for a recent method and reference to other work). These methods commonly employ a gas bubble of about 0·03 ml. with a 1 ml. sample. Both equilibrium and absorption of the gases during the analysis are carried out in the barrel of a Roughton-Scholander syringe (1943). This syringe has a fine-bore calibrated capillary (fused on to the barrel in place of the original nozzle) in which the bubble length can be measured. The bubble remains at atmospheric pressure during the equilibration process and in consequence these methods should not be applied, for reasons which are discussed in the next section, to samples in which the total tension of dissolved gases differs from atmospheric pressure. It would appear that this limitation has been overlooked in recent years except by Roos & Black (1950). Furthermore, no attempt has been made to allow for the influence of the partial pressures in the original bubble on the final state of equilibrium. Rather has the range of application of the methods been restricted to samples whose gas tension would not be significantly altered by equilibration with a bubble of alveolar air or artificial gas mixture of fixed composition. This restriction in the case of bloods with high oxygen capacities only rules out samples in which the pigment is practically fully saturated; then the oxygen buffering power of the pigment is nil. In the case of samples lacking an oxygen buffering system such restriction of the range of application would be a severe disadvantage. Given a large enough sample/bubble ratio the buffering of carbon dioxide tension changes is never a problem because of the high solubility of this gas.

The present paper describes a tonometric method for samples of 0·3 ml. or less, giving a degree of accuracy comparable to that of micro-Winkler methods for dissolved oxygen and requiring 20·25 min. for each determination. A readily available gas mixture of constant composition (atmospheric air) is used for the bubble and the influence of the latter on the equilibrium tensions can be precisely calculated and allowed for. It is applicable to both water and blood samples in which the total tension of dissolved gases is equal to or less than atmospheric pressure, and since oxygen, carbon dioxide and nitrogen can be determined on a single sample an estimate of total tension can readily be made.

This method has the wide utility of Krogh’s method (1908 a) but uses much smaller samples. Krogh allowed a stream of the sample to pass over the bubble (volume 0·004 ml.), whose pressure was adjusted until constant volume was obtained. This required at least 15 ml. of sample, but eliminated any influence of the initial bubble partial pressures on the final equilibrium. The present method more closely resembles that of Roos & Black (1950) in its manner of equalizing total pressure and total tension, but their method uses a sample of 17 ml. and by not allowing for the influence of the bubble its usefulness is restricted for the reasons discussed above.

A small bubble of atmospheric air is enclosed with the sample in a syringe and allowed to come into diffusion equilibrium with the dissolved gases. The bubble is then transferred to the analyser where its length is measured in a uniform fine-bore capillary before and after absorption of the carbon dioxide and the oxygen by the appropriate reagents. The equilibrium partial pressures and tensions differ to a greater or lesser degree from the original tensions in the sample according to the extent of the influence of the bubble, but the initial partial tensions can be calculated from the final partial pressures in the bubble.

The micro-analysis of the equilibrated bubble, using the method of Krogh (19086), presents no difficulties once the simple technique has been mastered. On the other hand, the establishment of a true equilibrium between sample and bubble calls for special care if the total tension of dissolved gases is less than atmospheric pressure. This is frequently the case in biological fluids and the extent of the ‘total tension deficiency’ may itself be of some interest.

Krogh (1908 a, 1913) has called attention to the necessity for adjusting the pressure of the gas phase to equal the total tension of the dissolved gases if a true diffusion equilibrium is to be reached. If the total tension of dissolved gases is lower than the total pressure of the gas phase, gas will be continuously given up to the liquid and the volume of the gas will steadily diminish until such time as the total tension becomes equal to the total pressure. If the system is open to the atmosphere the equilibrium total tension will be equal to atmospheric pressure (i.e. Equid saturated with gas) and a relatively small bubble may well disappear altogether before equilibrium is reached. At any intermediate stage before equilibrium is reached there will be a discrepancy between the partial tension and the partial pressure of each gas. This discrepancy will be greatest for the gas with the lowest diffusion velocity.

On the other hand, if a relatively small bubble is introduced into the liquid and the system then sealed the total pressure will be quickly adjusted to parity with the total tension because the bubble-volume must remain constant. A true diffusion equilibrium is attained very rapidly with small bubbles (see discussion on ‘specific surfaces’—Krogh, 1908a). On breaking the seal the total pressure of the bubble will return to atmospheric pressure and an instantaneous shrinkage will occur. This shrinkage will be proportional to the difference between the total tension and atmospheric pressure. Thereafter the total tension deficiency will, of course, cause an absorption of gas by the liquid as described above.

In the present method an air bubble of standard volume at atmospheric pressure is introduced into a syringe with the sample. The syringe is immediately sealed so that a true diffusion equilibrium can be reached, with the gas phase remaining at the standard volume. Afterwards the seal is broken, and before any appreciable further gas absorption can take place, the shrunken bubble is quickly transferred to the capillary of Krogh’s micro-gas-analyser (19086) and its reduced volume determined. The analysis then proceeds in the normal way by successive absorption of carbon dioxide and oxygen so that, the relative proportions of these gases having been determined, the individual partial pressures can be calculated.

As Krogh pointed out (1908 a), if the partial pressure of carbon dioxide only is required it is sufficient to equilibrate the bubble with the sample without adjusting the total pressure (or in the present method without sealing the syringe).

From a practical point of view there are considerable advantages in transferring the equilibrium bubble to a separate analyser as in the present method and in that of Roos & Black (1950). The danger of blood clots forming in the narrow analyser capillary (diameter 0·2 mm. in the present case) and the consequent necessity for the use of anti-coagulents are eliminated. Uniform drainage of the walls of the capillary, a factor which is vital to the accurate determination of the bubble volumes, is much more assured if the calibrated part is always wetted by the same solution. The latter circumstance also reduces the possibility of gas being given up to the bubble as a result of mixing air-saturated solutions of different concentrations (Krogh, 19086). The amount of gas exchange during the transfer has been found to be negligible.

Equilibration of the air bubble with the sample is carried out in an ‘Agla ‘micrometer syringe which has been modified as follows. The rather wide-bore glass nozzle has been replaced by one with a bore of 0-25 mm. At the inner end of this capillary nozzle a slight hollow is ground to assist in positioning the bubble below the capillary for speedy ejection from the syringe. With this capillary the dead space of the syringe is reduced to less than 0-002 ml. A 45° shoulder is ground on the tip of the nozzle, while behind this the nozzle is ground to fit a Record hypodermic needle mount. A flat brass disk carrying a pair of small hooks is cemented to the head of the syringe plunger, so that a rubber band fastened to the micrometer barrel can hold the head of the plunger firmly against the end of the micrometer screw at all times. The barrel of the syringe is clamped to the micrometer in such a position that the scale reading with the plunger right home is exactly 5 mm. This gives adequate room for applying the ‘Apiezon’ seal as described below. As the ‘Agla’ syringe is normally supplied^* with a plunger just slightly too short to reach the bottom of the barrel, it is necessary to grind a little off the rim of the open end of the barrel.

After the sample and the bubble have been introduced the syringe nozzle is sealed by applying a drop of hot paraflin wax by means of a wire loop just large enough to fit on to the shoulder of the nozzle tip. The space between the plunger and the end of the syringe barrel is sealed by a liberal application of hot ‘Apiezon’ grease. This is applied, by means of a fine glass jet hot enough to run immediately round the groove, as shown in Fig. 1 a, without further heating. Wax seals are unsatisfactory in this position because they have been found to contract on setting, causing a reduction of pressure within the syringe. Furthermore, bubbles of gas tend to appear the interface between the wax and the water film between plunger and barrel; these must, by their expansion, bias the equilibrium between the sample and the definitive bubble. The micrometer-syringe is mounted so that it can be rotated end-over-end by a small electric motor (see Fig. 1 b).

The analysis of the equilibrated bubble in the micro-gas-analyser follows Krogh’s procedure (19086) fairly closely, though several points of detail have been changed. The apparatus is filled with 1 % hydrochloric acid (instead of water) in order to ensure that there are no traces of alkali which could cause premature absorption of carbon dioxide. The adjusting screw and the boss in which it works are made of ‘Perspex’ instead of iron, to prevent any reaction with the hydrochloric acid. Finally, the alkaline pyrogallol solution for oxygen absorption has been replaced by the more satisfactory alkaline hydrosulphite solution (100 g. sodium hydrosulphite and 10 g. sodium anthraquinone-β-sulphonate are kept as a dry stock mixture ; 5 g. portions of this mixture are made up to 25 ml. in normal sodium hydroxide quickly filtered through cotton wool and stored under a layer of liquid paraffin).

The complete apparatus and the reagent bottles (1% hydrochloric acid, 10% sodium hydroxide and oxygen absorbent) are most conveniently mounted on a single 3 ft. retort stand as shown in Fig. 1b. At the lower end is mounted horizontally a bicycle hub whose spindle carries a pulley at one end and a clamp at the other. The complete micrometer syringe is held in the clamp so that it is roughly balanced on the spindle and can rotate end-over-end in a vertical plane at right angles to the operator’s line of vision. The pulley is driven by a small rheostat-controlled motor, the speed being adjusted so that the bubble oscillates fairly rapidly from end to end of the sample in the syringe.

The analyser is mounted above at such a level that the bottom of the cup just clears the tip of the nozzle when the syringe is rotated to the vertical position. A retort clamp which grips the water-jacket of the analyser is conveniently held by a friction-grip sleeve in a retort-stand boss so that the rotation of the analyser in the vertical plane can be quickly and precisely adjusted with one hand. At the top of the stand the various reagent bottles with siphon outlets are clamped. A fine glass jet connected to a suction line is used to empty the cup of the analyser and this with its collection trap is mounted just below the reagent bottles.

Since the success of the method depends largely on consistent attention to a number of points of detail in the manipulation, the detailed procedure of atypical determination is given below. The apparatus and reagents having been set up as described in the previous section and the analyser having been filled with hydrochloric acid to the exclusion of all air bubbles, proceed as follows:

1. Fill the dead space of the syringe with distilled water and screw down the plunger to the bottom of the barrel (micrometer reading = 5 mm.). Since the dead space volume is only 0-5% of the sample volume, the inclusion of the dissolved gases in this small amount of water may be disregarded. With the syringe in the horizontal position touch off the last drop of expelled water with filter-paper.

2. Couple the syringe to the sample container without intervening air bubbles and draw back the plunger until the micrometer reads 20-1 mm.

3. Disconnect the sample container, rotate the syringe to the vertical position (nozzle uppermost), adjust the micrometer to 20-0 mm. and touch off the drop with filter-paper.

4. Adjust the micrometer to 20-25 mm. and then to 20-15 mm., thus drawing in exactly 0-003 ml-of air.

5. On the flat end of the nozzle place a small drop of water to seal the bubble in the capillary and adjust the micrometer until the bubble is just drawn into the barrel of the syringe. Displace the bubble to one side of the nozzle and readjust the micrometer to 20·15 mm.

6. Dry the nozzle tip and then adjust the micrometer until the water forms a tiny convex meniscus at the end of the nozzle capillary.

7. Immediately apply a drop of molten paraffin wax to the end of the nozzle by means of the wire loop.

8. Heat a glass dropper containing ‘Apiezon’ grease so that on rotating the syringe through 180° and applying the grease, the latter runs smoothly round the groove between the syringe barrel and the plunger. Enough grease should be applied to form a seal of the shape shown in Fig. 1 a.

9. Allow time for the grease to harden and if necessary dry outside of the syringe with filter-paper.

10. Switch on the motor and allow the syringe to rotate end-over-end for 10 min.

11. Switch off the motor, bring the nozzle uppermost and register the bubble in the hollow at the bottom of the nozzle capillary.

12. Fill the cup of the analyser with a portion of the original sample (about 0·1 ml. is required). This should be omitted if the sample is a blood which clots readily and may be omitted otherwise unless the maximum accuracy is required with samples containing high partial tensions of carbon dioxide. In such cases the cup is filled with 1 % hydrochloric acid.

13. Without jolting the bubble out of the hollow, strike off the seal at the tip of the nozzle with the point of a needle and carefully expel the air drawn into the capillary by the contraction of the equilibrated bubble.

14. Immerse the tip of the nozzle in the cup of the analyser and as quickly as possible expel the bubble. Draw the bubble into the capillary immediately. (Avoid getting fragments of wax into the cup.)

15. Carry out the analysis of the bubble in the normal way.

16. Before commencing the next determination clean the grease off the plunger and the barrel with filter-paper and xylol.

During the gas absorption stages of the analysis the bubble should remain in the cup only so long as is necessary for complete absorption, otherwise an appreciable amount of gas exchange may occur between the bubble and the reagents. If the amount of carbon dioxide is known to be negligible the bubble can be exposed to the oxygen absorbent directly after measuring its initial length; but if carbon dioxide is not negligible it must first be absorbed separately by exposure to 10% sodium hydroxide in the cup. Each determination then yields three measurements of the length of the bubble : before and after absorption of carbon dioxide and after absorption of oxygen. The temperature of the sample when the bubble is introduced the temperature of the analyser water-jacket at the time of each measurement of bubble length and the barometric pressure should all be noted. The treatment of these data is discussed in the next section.

When the total tension of dissolved gases and the individual partial tensions are close to the values for atmospheric air, it may be sufficient to deduce the total tension directly from the degree of shrinkage of the standard bubble due to equilibration, and the partial tensions from the further shrinkages due to absorption of carbon dioxide and oxygen, respectively. In most cases, however, where a reasonable degree of accuracy is required, it is necessary to take account of the influence of the bubble itself on the final state of equilibrium. By a simple algebraic treatment of the system it is possible to derive a formula (which contains the necessary correction factor) for the calculation of the partial tension of the dissolved gas.

By means of a calibration table or conversion factor, bubble lengths measured in the analyser capillary are converted to volumes. The following symbols are used :

The above factors are measured with each determination.

The following treatment illustrates the manner in which the partial tension $(tO2)$ of oxygen in the original sample should be calculated. In the closed system the amount of oxygen is constant though the distribution of the gas between the gas and liquid phases alters during the equilibration. The states of partition before and after equilibration may be equated, the volumes of oxygen being measured or calculated at B mm. and T°A.

After equilibration:

The first term in this equation represents the simple proportionality between total and partial pressures and volumes in the equilibrated bubble, while the second term is the correction factor for the influence of the bubble on the final equilibrium. For the calculation of the constant k₂ the values α₂B and T need to be known approximately. The value of the absorption coefficient α₂, which is constant for one particular kind of sample at a given working temperature, may be taken as the same as for water in the case of dilute aqueous solutions. For the body fluids of most animals the value of equimolar sea water will suffice, while for other kinds of samples the solubility can, if necessary, be determined with the van Slyke apparatus.

If the sample is a blood containing a respiratory pigment the amount of contained gas will not be directly pressure-dependent. However, if the dissociation data and the oxygen capacity are known it is possible to draw the dissociation curve in the form relating volumes per cent, of combined oxygen to oxygen partial tension. If the approximate oxygen partial tension is calculated a straight line can be drawn tangential to the dissociation curve at the relevant partial pressure. By extrapolating this line to o and to 760 mm. the appropriate value of α₂ can be determined graphically. In many cases the value of thus deduced will give a value of k₂ so large as to make the correction factor insignificant, as it is in the case of carbon dioxide (see below). The principal value of this procedure will be in dealing with blood samples which are almost fully saturated with oxygen or have very low oxygen capacities. When the pigment is fully saturated combined oxygen is constant and only α₂ for dissolved gas need be considered. For pigments which are nearly saturated or have very low oxygen capacities the plasma oxygen may be a significant part of the pressure-dependent total, in which case the value of the absorption coefficient for dissolved oxygen should be added to the value deduced for the pigment by the above procedure.

Without sensible error T and B may be given fixed values in calculating k₂ for a day’s working, the error due to any normal variation in room temperature or barometric pressure being of no practical significance in this context.

The above formula serves equally well for the calculation of the partial pressure of carbon dioxide if the appropriate substitutions are made. Thus V is substituted for V₁, V₁for V₂, for k₂ and o for 0·00062 (carbon dioxide in the initial bubble can be neglected). However, because of the high solubility of carbon dioxide, the influence of the bubble on the equilibrium partial tension of the gas is very small. The correction factor has a constant value (if the carbon dioxide in the initial bubble of atmospheric air is neglected) of + 1·1 % of the value obtained from the expression (V − V₁)(B− p)/0·003. If the sample contains a carbon dioxide buffering system the correction factor will be quite insignificant.

The shrinkage of the equilibrated bubble after breaking the seal is proportional to the ratio of atmospheric pressure to the equilibrium total pressure (= total tension); but the latter may differ significantly from the original total tension of gases in the sample because of the influence of the gases in the initial bubble on the final equilibrium. In such cases the degree of shrinkage does not give an accurate estimate of the total tension of gases in the original sample. However, if a value for the total tension is required, it is a simple matter to repeat the above calculation for the nitrogen in the sample. The value of k₃ should be calculated and the appropriate values substituted for the volume of nitrogen in the initial bubble (0·00238 ml.) and in the equilibrated bubble (Ig). The total tension of dissolved gases is, of course, the sum of the calculated partial tensions of carbon dioxide, oxygen and nitrogen.

Before carrying out the calculations described above, the influence of any temperature variation on the measured volumes of the bubble must be allowed for. The analyser capillary should be calibrated with reference to a standard bubble measured by the micrometer syringe at the same temperature as the water-jacket of the analyser. Subsequently, for each determination four temperatures should be recorded : the temperature of the syringe and sample when the bubble is introduced (= temperature of the room air if this is not fluctuating too rapidly) and the temperature of the water-jacket at the three times of measuring the bubble in the capillary. If any of the last three differ from the first, appropriate corrections should be applied to V, V₁ and V₂. This is conveniently done by adding (subtracting) to the logarithm of the observed volume 0·00015 for every 0·1° C. fall (rise) in temperature.

In theory, account should be taken of any variation in the temperature of the sample during the equilibration period. This will alter the absorption coefficients and so cause a change in the partial tensions of the dissolved gases. At normal room temperatures a T C. change will cause an approximate change of 2·8, 1·9 and 1·5 % in the partial tensions of dissolved carbon dioxide, oxygen and nitrogen, respectively. It has been found that with reasonable precautions the temperature of the sample during the equilibration period can be kept constant within ± 0·2° C. without recourse to a water-bath. If the method is to be used with samples containing a respiratory pigment whose oxygen capacity has a high temperature coefficient more stringent temperature control may be necessary.

The procedure set out above has been devised to minimize a number of sources of error. Foremost among these is the excessive shrinkage of the bubble in the unsealed syringe if the total tension of dissolved gases is less than atmospheric pressure. The time during which the air bubble lies in contact with sample in the unsealed syringe or in the analyser cup must therefore be kept as short as possible and agitation kept to a minimum. The accuracy with which the standard bubble is measured and drawn into the syringe by the micrometer and the efficacy of the seals are also important factors. The standard bubble can be reproduced with an accuracy of + 1 % volume if the micrometer is carefully used and the capillary of the syringe nozzle is kept clean. It has been found that the sealing technique as described is entirely adequate, apart from an occasional failure of the nozzle seal. This fault reveals itself by a leakage of air into the capillary of the nozzle before the seal is struck off at the end of the equilibration period. It should be a matter of routine to check this point after every equilibration.

For a discussion of factors affecting the accuracy of the analysis of the bubble reference should be made to Krogh’s original paper (19086). It may be remarked that in the present method the pre-analysis stages in the manipulations are likely to contain the more important sources of error (excluding errors of meniscus readings). Adherence to the details of the procedure set out above and attention to the cleanliness of the syringe (especially of the capillary nozzle) will ensure a standard of accuracy consistent with the examples set out in Table 1.

Distilled water samples were equilibrated with a variety of mixtures of carbon dioxide, oxygen and nitrogen at total pressures ranging from 750 down to 510 mm. Three portions of each were subjected to complete analysis by the tonometric method, and duplicate analyses, for dissolved oxygen only, were also made on additional portions by the micro-Winkler method of Fox & Wingfield (1938). The results of these determinations are set out in Table 1. The tonometric determinations show a high degree of consistency as indicated by the coefficients of variability (column headed F). On the basis of these values of V, replicate determinations of carbon dioxide, oxygen and total gas tension are seen to agree within 3·0, 2·2 and 1·2 %, respectively. The tendency for V to increase as the partial tensions (= volume differences) decrease is due, in part at least, to the fact that the possible error in reading the meniscus at each end of the bubble is absolute and assumes a greater importance as the volumes (V ‒V₁) and (V₁ ‒V₂) decrease. Each calculated partial tension is based on two volume measurements, i.e. four meniscus readings. In addition, the values of V will be influenced by variations in the volume of the standard bubble, and by the extent of unwanted gas exchange between bubble and sample in the unsealed syringe. The latter factor will be of greatest importance when the oxygen tension is low and the carbon dioxide tension high. Because of differences in diffusion velocities these unwanted gas exchanges are more serious in the case of carbon dioxide than of oxygen or nitrogen. This is reflected in the higher values of V obtained with carbon dioxide determinations. Values of V for total gas tensions (sum of separately determined carbon dioxide, oxygen and nitrogen tensions) are consistently lower than the values for carbon dioxide and oxygen, because the determination of the large proportion of nitrogen is less influenced by the factors mentioned above; the effect of unwanted gas exchange is negligible in the case of nitrogen and the errors of meniscus readings are relatively very small, leaving variation in the volume of the standard bubble as the principal source of error. The actual values of V for the nitrogen determinations (not included in the table) range from o to 1·28. It should be noted that for partial pressures substantially less than 10 mm. the errors involved in reading the meniscus of the bubble are liable to increase the overall error of the determination beyond the limits stated above. The values of 1·0 and 1·5 mm. for carbon dioxide in sample 1 represent barely detectable differences between V and V₁.

Finally, comparison of tonometric and Winkler determinations of oxygen tension shows a close agreement between the means obtained by the two methods. The maximum difference is one of 5·4% in the case of sample 4, while the mean percentage difference is 3·0. In all but two of the samples, in which the difference between the results of the two methods is insignificant, the Winkler method gives slightly higher results than the tonometric method.

I wish to thank Prof. C. J. van de Klaauw and Prof. D. J. Kuenen of the Zoologisch Laboratorium der Rijksuniversiteit te Leiden for their kindness in permitting me to work for 2 weeks in their respective institute and department, where this work was begun. To Dr H. P. Wolvekamp, now of the Instituut voor Experimentele Dierkunde, Leiden, I am especially indebted for the suggestion of equilibrating air bubbles with samples in a closed system and for much helpful discussion and encouragement of my first attempts to put the idea into practice in his laboratory. I also wish to express my sincere appreciation of his hospitality and kindness throughout my stay in Leiden. My thanks are also due to my colleague Dr F. Segrove for his helpful criticism of this account and to Prof. Q. H. Gibson, Department of Biochemistry, University of Sheffield for discussion of a number of theoretical points. The expenses of the visit to Leiden were largely met by a grant from the University of Sheffield Foreign Travel Fund.