Əsas səhifə
Biotechnology and Bioengineering Reaction kinetics in biofilms
Reaction kinetics in biofilms
Zbigniw Lewandowski, Gabriele Walser, William G. CharacklisBu kitabı nə dərəcədə bəyəndiniz?
Yüklənmiş faylın keyfiyyəti necədir?
Kitabın keyfiyyətini qiymətləndirə bilmək üçün onu yükləyin
Yüklənmiş faylların keyfiyyəti necədir?
Tom:
38
İl:
1991
Dil:
english
Səhifələr:
6
DOI:
10.1002/bit.260380809
Fayl:
PDF, 1,76 MB
Sizin teqləriniz:
 Əvvəlcə öz hesabınıza daxil olmağınız xahiş olunur

Kömək lazımdır? Kitabı kindleyə necə göndərməli qısa təlimatını oxumağınız xahiş olunur
15 dəqiqə ərzində epoçtunuz bərpa olunacadır.
15 dəqiqə ərzində Kindle hesabınız bərpa olunacadır.
Qeyd: Kindleyə göndərdiyiniz hər bir kitabı verifikasiya etməlisiniz. Amazon Kindle Supportdan məktub alıbalmadığınıza dair elektron poçt ünvanınızı yoxlayın.
Qeyd: Kindleyə göndərdiyiniz hər bir kitabı verifikasiya etməlisiniz. Amazon Kindle Supportdan məktub alıbalmadığınıza dair elektron poçt ünvanınızı yoxlayın.
Əlaqəli Kitab siyahıları
0 comments
Kitab haqqında rəy bildirə və öz təcrübənizi bölüşə bilərsiniz. Digər oxuculara Sizin oxuduğunuz kitablar haqqında fikrinizi bilmək maraqlıdır. Kitabın ürəyinizcə olubolmamasından asılı olmayaraq, bu barədə dürüst və ətraflı məlumat versəniz, insanlar onlar üçün maraqlı olan yeni kitablar tapa bilərlər.
1


2


Reaction Kinetics in Biofilms Zbigniew Lewandowski,* Gabriele Walser, and William G. Characklis National Science Foundation Research Center for Interfacial Microbial Process Engineering, Montana State Unversity, Bozeman, Montana 59717 Received August 2, 1990!4 ccepted February 25, 1991 A novel in situ microtechniqueallows evaluating parameters of diffusioncontrolled reactions in biofilms. A microprobe, 15pm in diameter, was used to simultaneously measure the dissolved oxygen concentration and the optical density at different depths in a submerged biofilm. Based on the results, the biofilm diffusion coefficient for dissolved oxygen, D,, the dissolved oxygen flux through the biofilm surface, Jo2,and the half velocity coefficient, K,, have been calculated. Key words: microtechnique microprobe biofilm dissolved oxygen concentration  INTRODUCTION Surfaces immersed in water frequently become colonized by microbial films held together by extracellular polymeric substances (EPS). Diffusivity through these biofilms controls the microbial metabolic reaction rates. For example, the microbial respiration reaction is diffusioncontrolled. This may have repercussions for biofilm management in many applications. Bacterial biofilm infections often do not respond to antibiotic treatment. It has been postulated that the extracellular polymers constitute a diffusion barrier for antibiotic^.^ An inhibition of antibiotic diffusion has been observed in the presence of alginate, a polyanionic polysaccharide, and in the presence of the polysaccharide from Pseudomonas aeruginosa. l7 Although new findings indicate more complex mechanisms for the biofilm resistance to antibiotic treatment, the penetration models inevitably contain diffusivity terms.’ In industry, the metabolic reactions mediated by microorganisms residing in biofilms promote the biodeterioration of materials including metals, concrete, and plastics. Microbially induced corrosion alone costs the United States economy billions of dollars every year.7 To evaluate ; the rate and extent of biofilm processes, the understanding of the growth and respiration kinetics is fundamental. Modeling diffusion with chemical reaction often complicates representation of the system parameters. This is especially true for the diffusion coefficient. The system itself is complex and exact mathematical solutions are not always available. Numerical procedures frequently use the diffusion coefficient as a control parameter to “adjust” or to “calibrate” the models to observed data. Such tuning, although effective for description of experimental results, causes understandable reservations regarding the predictive value of the * To whom all correspondence should be addressed. Biotechnology and Bioengineering, Vol. 38, Pp. 877882 (1991) 0 1991 John Wiley & Sons, Inc. model. An in situ method is needed for evaluation of diffusioncontrolled reaction parameters in thin layers containing microbial activity. Techniques for determining biofilm reaction kinetics and the related diffusion coefficient depend on two types of analyses: (1) chemical analysis of bulk water, and (2) measurements inside the biofilm using microsensors. The techniques based on chemical analysis of bulk water can be applied when the biofilm is growing at a water permeable membrane. The diffusion coefficient of dissolved oxygen (DO) through an artifically induced biofilm formed by filtering nitrifiers through a membrane filter, was found between 80 and 100% of its value in water.’l The dissolved oxygen diffusion coefficient through an artifically induced biofilm formed by filtering an activated sludge was calculated between 20 and 100% of that in water.’ The same coefficient for a heterotrophic biofilm growing on a membrane filter was found between 40 and 140% of its value in water.I6 The increase in diffusivity over that in water was ascribed to eddy diffusion in the biofilm matrix close to the biofilmwater interface. Microsensors, usually in the form of microelectrodes, are increasingly popular for biofilm research.Zv’03’2320 The microelectrode technique, using a stationary microelectode placed in the biofilm, has been applied to dissolved oxygen diffusivity mea~urements.~”~ This procedure assumed that respiration reaction in the film was zero order, simplifying the diffusion equation to obtain an exact solution. The diffusion coefficient for dissolved oxygen through a slime originating from trickcm2 s’. ling filters by this method was Of= 4.0 x Another procedure to measure the dissolved oxygen diffusion coefficient using a stationary microelectrode required incubation of sediments for three weeks in O.1M solution of HgClz to kill the microorganisms before the measurement.” The technique eliminated reaction from the diffusion equation assuming that the treatment did not change sediment properties. The measured diffusion coefficient of dissolved oxygen through sediments was 1.4 x cm2 s’. All of the above techniques require extensive sample preparation and often involve simplifying assumptions to calculate the coefficient. Theoretical models predict biofilm system behavior but can only be experimentally verified if satisfactory experimental procedures exist. The microsensor technique is the most promising approach to supply the necessary information. 13~1491’~19 CCC 00063592/91/08087706$04.00 Uncertainty in locating the biofilmwater interface presents a serious difficulty in relating observations to theoretical calculations. Since the reactions in biofilm are diffusioncontrolled, the results of microelectrode measurements constitute concentration profiles. In effect, the microelectrode measurement yields a concentration as a function of distance in the biofilm. However, there is no reported procedure for locating the biofilmwater interface in these profiles. This article describes an independent measurement for locating the biofilmwater interface. The biofilm is penetrated with a microprobe, 15 pm in diameter, simultaneously measuring substrate concentration and optical density. The biofilmwater interface is identified from the optical density profile. Superimposing the biofilmwater interface on the substrate concentration profile enables direct comparison of theoretical models to experimental results. EXPERIMENTAL METHODS The dissolved oxygen microelectrode (tip diameter of 5 pm)was fabricated as described previously.6The electrode was calibrated in airsaturated and subsequently deoxygenated water. The effect of stirring' was less than 5% of the response. A singlecable, glass fiber optic 125/85 cable (United Detector Technology) was used to fabricate the optical density sensor. Three centimeters of the fiber tip was mechanically exposed. The exposed fiber was first heated and pulled, then cut. The tip was etched in hydrofluoric acid (HF) to obtain a sharp tip of 10 pm in diameter. Etching without pulling, which resulted in thicker sensor tips, was also attempted. The fabrication procedures did not influence the results. The light source was a lightemitting diode (LED), 940nm high radiant emitter, model HEMT3301 (HewlettPackard). A squarewave, frequencymodulated 500Hz light produced by a custom generator was used. The detector was a standard highresponsivity photodetector, model PINHR020 (United Detector Technology). The DO microelectrode was coupled with the fiber optrode to form a dual sensor microprobe. The coupling was performed under a microscope equipped with a video camera so that the manipulations were observed on the screen. A fastreacting epoxy glue was used to fuse the sensors together. Care was exercised to place the sensors tips in adjacent positions so they would be at the same level in the biofilm (Fig. 1). A mixed population biofilm was accumulated in a rotating biological contactor (RBC) using a feed solution of glucose and mineral salts. The rotating polycarbonate discs were equipped with removable slides. A slide was removed from the reactor, cleaned on one side, and placed in a Petri dish with the biofilm up. It was important to place the sample on a transparent sublayer. The Petri dish was filled with the feeding solution and 878 OPT1 C A L D E N S I TY SENSOR Figure 1. Photomicrograph of the tip of the microprobe. The tip diameter is 15 urn micrometer at the marked location. placed on a stand equipped with a microscope stage. The solution was continuously aerated (and mixed) using a glass capillary immersed in the liquid. The probe was mounted in a motorized micromanipulator and positioned directly above the film. The LED was mounted below the Petri dish in a handoperated micromanipulator (World Precision Instruments, Inc.). The radiation intensity was monitored using a voltmeter. The microprobe above the film and LED below the film were horizontally manipulated to obtain the maximum incident radiation to align the probe and the light source. Microelectrode measurements are susceptible to electromagnetic noise. To protect the system, the measurement and the data acquisition were totally automatic. The measurement was performed using a dedicated electrical circuit in the building. The operator left the room after initiating the microprobe movement to reduce electromagnetic noise. The probe penetrated the sample at a speed of 2.5 pm/s for a 2.54 interval alternating with a 2.5s stationary period, during which the measurements of light intensity and dissolved oxygen concentration were conducted. The instruments were linked with a data acquisition system and the profiles of dissolved oxygen and light absorption were simultaneously presented on the screen and recorded (Fig. 2). Figure 2. The experimental system for determining the dissolved oxygen and optical density profiles in a biofilm. BIOTECHNOLOGY AND BIOENGINEERING, VOL. 38, NO. 8, OCTOBER 20, 1991 INTERPRETATION OF RESULTS The biofilm and substratum surfaces were identified by measured changes in optical density. The biofilmwater interface and the substratumbiofilm interface could be located in the dissolved oxygen profile since the tips of the sensors were level (Fig. 1). Two abrupt changes in the light intensity profiles occurred (Fig. 3). The first (marked “1”) occurs where the probe enters the biofilm; the second (marked “2”) occurs where the probe touches the substratum. When the probe enters the biofilm, a change in optical density is observed. When the probe touches the substratum, the fiberoptic sensor bends slightly, which results in a decrease in light intensity. Point “2” was identified as the biofilmsubstratum interface and the light intensity monitored at this place was accepted as the incident radiation intensity for further calculation. The results of the optical density measurements (Fig. 3) were expressed according to Lambert’s law: log(Zo/Z) = EL, where Z, is the intensity of the radiation incident (maximum radiation monitored at the “Y, Fig. 3); Z is the emergent intensity; E is the extinction coefficient; and L is the layer thickness. The calculated log(Z,/Z) values are a function of distance (L).3The dissovled oxygen concentration was calculated from a microelectrode calibration curve. RESULTS Results of measurements, interpreted and calculated according to the above procedures, are presented in Figure 4.Two profiles are presented; this of dissolved oxygen concentration and that of optical density. The profile of optical density indicates constant extinction coefficient in water. Entering the biofilm by the microprobe is marked as a sudden change in extinction coefficient. Extinction coefficient across the biofilm is also constant. The dissolved oxygen concentration decreased 3 W f substratum surface 2n v) W a i 2z W I 2a 0 z w 3 0  0 2 2 g v, n 6.0 0.21 5.0 0.18 0 = u 40 0.15 3.0 0.12 I 2.0 0.09 1.0 0.06 0.0 0.03 1.0 0.00 0.30 0.60 0.90 1.20 1.50 1.80 2.10 0 0.00 2.40 DISTANCE, mm Figure 4. The variation of dissolved oxygen and absorbance with distance in a biofilm system. The abrupt change in the slope of absorbance profile defines the biofilmwater interface. from 8 mg L’ in bulk water to zero in the biofilm. Superimposing the optical density profile on the dissolved oxygen concentration profile permits further estimations. T h e biofilmwater interface was positioned 0.7 mm from the substratum. Dissolved oxygen concentration at the biofilmwater interface was determined to be 1.2 mg LI. The mass transfer boundary layer thickness was measured as 1.33 mm and the decrease in dissolved oxygen concentration within this layer was calculated as 6.8 mg L’. Locating the position of the biofilm surface at the dissolved oxygen profile permitted separation of the dissolved oxygen concentration profile into two parts, that in the water (diffusion layer) and that in the biofilm. THEORETICAL CONSIDERATIONS The rate of change of substrate concentration in the biofilm can be described by an equation resulting from a differential mass balance: where Dfis the diffusion coefficient for the dissolved oxygen in the biofilm (cm2s’); C i s the dissolved oxygen concentration at a point x (mg LI); V,,, (mg L’ sI) and K, (mg LI) have the usual meaning in the MichaelisMenten equation. T h e steadystate concentration within a biofilm (dC/dt = O)/ is achieved when consumption equals the rate of transport due to diffusion: biofilm surface IU U & 1 sensor movement DISTANCE, micrometers Figure 3. The change in current and potential as a function of distance permit location of the biofilmwater and biofilmsubstratum interfaces. Equation (2) cannot be solved exactly but it can be integrated once.’ Using this technique, the integration constant can be found from the boundary conditions. If the film is not totally penetrated by the constituent, then ( d C / d ~=) ~0 for C = 0 (no mass transport beyond this depth). If the biofilm is totally penetrated and the concentration at the substratum surface equals C , , then LEWANDOWSKI, WALSER, AND CHARACKLIS: REACTION KINETICS IN BlOFlLMS a79 (dC/dx)f = 0 for C = C, (no penetration at the wall). For partially penetratd biofilms, Halfsaturation Coefficient, K, For a totally penetrated biofilm: The slope divided by the intercept in the plot (Fig. 6) of the inverse of the second derivative against the inverse of the concentration gives K,. The second derivative was evaluated based on the polynomial curvefitting equation (Fig. 5). (g),  C,  K, * In Ks = d2%(c K, “) + C, + (4) The reaction rate ( R ) is equal to the rate of mass transfer across the biofilmwater interface, and can be directly calculated from the above equations as R =A * Df(2) (5) f where A is the area of the biofilmwater interface. Equation (4)can then be considered a general form of eq. (3). Partially penetrated biofilms can be considered a specific case of a totally penetrated biofilm. The reaction rate at the biofilmwater interface can be described as: C ,  C,  K, * ln Ks + “’) Ks + C , Inversion of eq. (2) permits estimation of the K , value: Diffusion Coefficient The biofilm is partially penetrated with oxygen (Fig. 4). Thus, the first derivative dC/dx should be linearly related to the {C  K, In[(C + K,)/K,])”*. The regression line (Fig. 7 ) is as follows: (s), = 75.984C  K, K, +C (8) ~ K, The derivative (dC/dx)f at the biofilmwater interface can be calculated from eq. (8). For the dissolved oxygen concentration at the biofilmwater interface, C = C , = 1.2 mg L’ and K , = 0.25 mg L’ (6) where C, is the biofilmwaterinterface substrate concentration andA is the biofilm surface area (cm2).When the biofilm is totally penetrated by substrate, C , = 0, eq. (6) is reduced to the form derived from eq. (3). * In ~ 0.002  0.0018 0.0016 0.0014 9 F 2 K, = 0.25mgll 0.00120.001 9 2 0.0008 CALCULATION OF KINETIC PARAMETERS 0.0006 The biofilmwater interface was located 0.07 cm above the substratum. The concentration profile (Fig. 4) can then be divided into two parts: that in the bulk water and that in the biofilm (Fig. 5). A thirdorder polynomial curvefitting procedure was used to estimate the observed dissolved oxygen profile (Fig. 5). BlOFlLM 002 003 005 0.000204 0 1 2 3 006 007 5 4 6 1IC Figure 6. Determination of the half saturation coefficient by linear regression [eq. ( 7 ) ] . ‘I 1 20 rngh  004 0.0004 dC/dX = 75.98 ( C  Ks Ln + ~ Ks C Ks ”* ) 008 DISTANCE, cm Figure 5. The dissolved oxygen profile within the biofilm was estimated by polynomial regression technique. 880 [C  Ks Ln ((C Figure 7. + Ks)/Ks)]’I2 Determination of 2Vm,,/D, [eq. (3)]. BIOTECHNOLOGY AND BIOENGINEERING, VOL. 38, NO. 8, OCTOBER 20, 1991 (dC/dx)f = 57.8 mg L’ cm’. T h e flux of oxygen through the biofilmwater interface is: JJ = D J ( Z ) f (9) where subscript “f” stands for “film.” The flux of oxygen through the diffusion layer is: J , = Dw($) where subscript “w” stands for “water.” The flux continuity must be preserved at the biofilmwater interface (Jf = J,): The derivative (dC/dx), calculated from the slope in Figure 4 equals 51 mg L’ cm’. The biofilm diffusion coefficient can be calculated from eq. (11): Substituting the calculated values, Df is estimated as 0.880,. Using the dissolved oxygen diffusion coefficient in water at 2loC,” D , = 2.0 x cm2 s’, the dissolved oxygen diffusion coefficient in the biofilm is Df = 1.76 x cm2 SKI. Oxygen Uptake Rate Diffusion coefficient permits further calculations. Based on eq. (lo), the dissolved oxygen flux through the biofilmwater interface is J , = 1.02 x mg s’ cm*. W,,, can be evaluated from Figure 7 and reaction rate from eq. (6). DISCUSSION AND CONCLUSIONS Advanced biofilm research depend upon sensor miniaturization. Biofilm thicknesses are typically in the range of between a few and a few hundred micrometers. Sensors, to effectively take measurements within biofilms without disturbing their structure, must be reduced to tip diameters of less than 10 pm. Suitable mathematical models are imperitive to interpret the results from microsensors measurements. An algorithm and instrumentation for measuring respiration reaction kinetics in biofilms has been presented. The microprobe simultaneously measures the oxygen concentration and optical density profiles across a biofilm. The optical density profile can be used to identify the position of the biofilmwater interface and the oxygen concentration profile yields the dissolved oxygen concentration at that interface. Based on the oxygen concentration profile and theoretical models, dissolved oxygen flux, diffusion coefficient and reaction rate in microbial films were estimated. This procedure is general and can be used for any substrate (e.g., organic compound, ion, or dissolved gas) for which a concentration profile across the biofilm can be measured. Availability of the specific microsensors is the obvious limitation to this procedure. The measurement is nondestructive and can be automated. A model describing biofilm respiration rates [eq. (6)] was used to describe biofilm activity as a function of oxygen concentration at the biofilmwater interface and biofilm substratum. The model does not require that bulk oxygen concentration be known. The bulk oxygen concentration and bulk hydraulic conditions influence the biofilm respiration rate only through the shape of the dissolved oxygen profile (C, and C o ) .All parameters of the model are experimentally accessible. Consequently, the model calculations can be then performed without “calibration” to improve the goodness of fit. It should be stressed that the measured and calculated parameters are site specific. That is, they apply only to the specific location in the biofilm penetrated by the sensor. Probes with small cross sections reduce the possibility of physical damages to the biofilm structure during penetration, but damage can occur unavoidably while withdrawing the probe from the biofilm. Since the probe can leave a pit at the site of penetration, the profile measurements can be made only once per site. Comparing the results from different locations to estimate the precision would require the assumption that the measured parameters had the same value. Such an assumption would not be warranted by this technique. For these reasons, it is hard to estimate the precision of individual measurements. Based on the algorithms described, the following parameters were calculated for oxygen in this particular film: diffusion coefficient for oxygen in the biofilm, Df = 1.76 x cm2 s’; dissolved oxygen flux at the biofilm water interface, Joz = 1.02 x mg s’ ern'; and the half saturation coefficient, K, = 0.25 mg L’. Respiration reaction rate coefficients in biofilms have been measured in laboratory systems by other methods. However, all previous procedures require extensive biofilm preparation or gross simplifying assumptions. The major contribution of this work is an integrated procedure using instrumentation and mathematical modeling to calculate the in situ kinetics of the biofilm reaction in an intact film. The measurement is nondestructive, automatic, and can be conducted in situ. The authors acknowledge their support from the Center for Interfacial Microbial Process Engineering at Montana State University, an Engineering Research Center sponsored by the National Science Foundation, and the Center’s Industrial Associates. LEWANDOWSKI, WALSER, AND CHARACKLIS: REACTION KINETICS IN BlOFlLMS 881 References 1. Baumgartl, H. 1987. Systematic Investigations of Needle Electrode Properties in Polarographic Measurements of Local Tissue POz. p. 1742. In: A.M. Ehrly, A. J. Hauss, R. Huch (ed.), Clinical Oxygen Pressure Measurement. Springer, Berlin. 2. Bungay, H.R., 111, Whalen, W. J., Sanders, W. M. 1969. Microprobe Techniques for Determining Diffusivities and Respiration Rates in Microbial Slime Systems. Biotechnol. Bioeng. 22: 765772. 3. Calder, A. B. 1969. Photometric Methods of Analysis. Adam Hilger Ltd: London. 4. Costerton, J.W., Cheng, K. J., Geesey, G . G . , Ladd, T. J., Nickel, J. C., Dasgupta, M., Marrie, T. J. 1987. Bacterial Biofilms in Nature and Disease. Ann. Rev. Microbiol. 41: 435464. 5. FrankKamenetskii, D. A. 1969. Diffusion and Heat Transfer in Chemical Kinetics. Plenum, New York. 6. Lewandowski, Z., Whon Chee Lee, Characklis, W. G., Little, B. 1989. Dissolved Oxygen and pH Microelectrode Measurements at Water Immersed Metal Surfaces. Corrosion 45: 9298. 7. Little, B. J., Wagner, P . A . , Gerchakov, S.M., Walch, M., Mitchell, R. 1986. The Involvement of a Thermophilic Bacterium in Corrosion Processes. Corrosion 42(9): 533536. 8. Matson, J.V., Characklis, W. G . 1976. Diffusion into Microbial Aggregates. Water Res. 10: 877885. 9. Nichols, W.W. 1989. Susceptibility of Biofilms to Toxic Compounds. p. 321331. In: W. G. Characklis and P. A. Wilderer (ed.), Structure and Function of Biofilms. Wiley. 10. Revsbech, N.P., Jorgensen, B. B. 1986. In: K . C . Marshall (ed.), Microelectrodes: Their Use in Microbial Ecology. Advances in Microbial Ecology. Plenum, New York. 11. Revsbech, N. P., Madsen, B., Jorgensen, B. B. 1986. Oxygen Production and Consumption in Sediments Determined at 882 12. 13. 14. 15. 16. 17, 18. 19. 20. 21. High Spatial Resolution by Computer Simulation of Oxygen Microelectrode Data. Limnol. Oceanogr. 31(2): 293304. Riethues, M., Buchholtz, R., Onken, O., Baumgartl, H., Lubbers, D.W. 1986. Determination of Oxygen Transfer from Single Air Bubbles to Liquids by Oxygen Microelectrodes. Chem. Eng. Process 20: 332337. Rittmann, B. E., McCarty, P. L. 1978. VariableOrder Model of BacterialFilm Kinetics. J. Environ. Eng. Div. EE5: 889900. Rittmann, B. E., McCarty, P. L. 1980. Model of Steady State Biofilm Kinetics. Biotechnol. Bioeng. 22: 23432357. Sanders, W. M., 111, Bungay, H. R., 111, Wahlen, W. J. 1970. Oxygen Microprobe Studies of Microbial Slime Films. Chem. Eng. Symp. Ser. Water 67: 6974. Siegrist, H., Gujer, W. 1985. Mass Transfer in a Heterotrophic Biofilm. Water Res. 11: 13691378. Slack, M. P. E., Nichols, W. W. 1981. The Penetration of Antibiotics through Sodium Alginate and through the Exopolysaccharides of a Mucoid Strain of Pseudornonas aeruginosa. Lancet ii:502503. Strand, S. E., McDonnel, A. J. 1985. Mathematical Analysis of Oxygen and Nitrate Consumption in Deep Microbial Films. Water Res. 19: 345352. Suidan, M.T., Wang, Y.T. Kim, B. R. 1989. Performance Evaluation of Biofilm Reactors Using Graphical Techniques. Water Res. 23: 837844. Whalen, W. J., Bungay, H .R., 111, Sanders, W. M., 111. 1969. Microelectrode Determination of Oxygen Profiles in Microbial Slime Systems. Environ. Sci. Technol. 3: 22972298. Williamson, K., McCarty, P . L . Verification Studies of the Biofilm Model for Bacterial Substrate Utilization. J. Water Pollut. Control Fed. 48: 281296. BIOTECHNOLOGY AND BIOENGINEERING, VOL. 38, NO. 8, OCTOBER 20, 1991