Because of recent events and the fact that there is so much misinformation being provided to the public from all sources, ASA suggested to Prof. Meselson that we reprint this article with an update from him to precede the reprint. The original ASA article received outstanding comments from scientists all over the world.

Note Regarding Source Strength
by Professor Matthew Meselson

The "Note Regarding Source Strength" reproduced below is the same as that published in the ASA Newsletter of June 8, 1995, except for the correction of a typographical error (the omission of "pi") in the equation for dose in Table 1.

          Note that source strength is defined as "the number of viable spores released at the source that travel in the atmosphere as particles small enough to initiate inhalation anthrax". Using this definition, the source strength estimates in Table IV are given in milligrams, taking the number of spores per milligram, as stated, as 10⁹. The question of whether the aerosol released at Sverdlovsk consisted only of viable spores or also contained inviable spores and other material is obviously not addressed in the present estimates. These estimates should be regarded only as what they are: estimates of source strength, as defined in the note, that follow from the stated assumptions regarding atmospheric dispersion and regarding dose-response relations for the infectious aerosol and the human population exposed to it.
           Although the present estimates follow from the assumptions made, the most relevant dose-response data available are for non-human primates, not for any human population, and none of it is for the low attack rates observed in the Sverdlovsk outbreak. Neither do we know if the virulence of anthrax spores in the aerosol released at Sverdlovsk was like that in aerosols employed in published experiments with monkeys. And even the well done experiments at Fort Detrick and Porton with monkeys gave ID₅₀ values covering a more than twenty-fold range -- from 2,000 to 45,000 respirable spores. These uncertainties are only imperfectly addressed by considering a number of different dose-response relations, as is done in Table IV.
            It may be of interest that the estimates presented here have an antecedent in an April 1980 memorandum I wrote for the US Interagency Sverdlovsk Working Group. We had no reliable information about the geographical distribution of attack rates. So we could only estimate what dose would be inhaled at a given downwind and crosswind distance from an aerosol release of a given number of spores under given atmospheric conditions. My estimates of dose as a function of source strength, based on a Gaussian plume model for a moderately stable atmosphere, were, as to be expected, in good agreement with those in the present Table II. Some members of the Working Group, apparently unfamiliar with atmospheric dispersion models, thought my estimates of source strength were far too low. The following month we received a memorandum from Dugway Proving Ground with estimates in essential agreement with those in my memorandum and in the present Table II. By then, however, a very much higher estimate had already been briefed to President Carter by Admiral Stansfield Turner, then Director of Central Intelligence.

(The reprinted article)
Note Regarding Source Strength
by Professor Matthew Meselson

           In November 1994, Prof. Meselson described the size and source of a plume of anthrax spores in Sverdlosk in 1979. The model employed and calculations performed to estimate the concentration of spores is presented here.
Cambridge, Massachusetts.
           This note provides background information regarding the estimate of Meselson et al. (Science 266: 1202-1208, 1994) that the aerosol that caused the Sverdlovsk anthrax outbreak of 1979 contained between a few milligrams and nearly a gram of anthrax spores. Figure and reference numbers correspond to those in the Science article.
            Estimating source strength. We define the source strength as the number of viable anthrax spores released at the source that travel in the atmosphere as particles small enough to initiate inhalation anthrax. Source strength may be estimated from knowledge of: (i) the attack rate among people at a particular location; (ii) the dose D (number of spores inhaled) corresponding to the attack rate, obtained from the dose-response relation; (iii) the ratio R of the dose at the particular location to the source strength, computed from the applicable atmospheric dispersion model and the breathing rate of the exposed population. The source strength is then D/R.
           Atmospheric dispersion model. The Gaussian plume model used in the article to compute dose as a function of source strength at any downwind distance (x) from the source and crosswind distance (y) from the cloud centerline is given in Table I. The model employs sigma-x and sigma-y values given by Briggs (21) for an atmosphere of neutral stability (stability "D") on open terrain. The breathing rate is taken as 30 L/min, as for a man engaged in light work (33).
            Comparison with other dispersion models. Table II gives centerline doses for a source strength of 10¹⁰ spores calculated from the atmospheric dispersion model and the breathing rate used in the article and given in Table I. For comparison, Table II also gives centerline doses calculated from three other atmospheric dispersion models: Gaussian plume with sigma values given by Briggs (21) for a slightly stable atmosphere (stability "E") on open terrain; POINT V Gaussian puff model for stability "D" with sigma values given in Table 2, page 17 in "Methodology for Chemical Hazard Prediction" (Department of Defense Explosives Safety Board, Washington DC, 1980); and the TNO "Yellow Book" Gaussian puff model for stability "D" (TNO Defence Research, Rijswijk, The Netherlands). For each model, it is assumed that there was no limit to vertical mixing. Depending on the actual height of the mixed layer, the tabulated doses may be subject to upward revision at large downwind distances.
            Attack rates. As described in the article, the attack rate at the ceramics factory, circa 2.8 km downwind of the source and apparently intersected by the cloud centerline, was about 2% (10/450) for pipe shop workers and 0.8% (8/1,050) for those who worked in buildings west of the pipe shop. The latter include anthrax patients who worked in the tile shop, maintenance shop, and cafeteria. The overall rate was therefor 1-2%. To calculate source strength, the article takes the attack rate 2.8 km directly downwind of the source as 2%.
           Attack rates may also be estimated for two other locations, not considered in the article. These are a residential section and an automotive maintenance and repair center located, respectively, upwind and downwind of the ceramics factory. The residential section had a population density of 10,000/km². The part of it within the outermost constant dosage contour of Figure 2 and bounded by the ceramics factory and Compound 32 covers 0.7 square kilometers. If 10% of the adult population were at home at the time of cloud passage (including absentees, night workers, pensioners, sick, unemployed, and vacationers), about 500 adults would have been present. Ten patients were mapped within the area in question (Figure 2), for an average attack rate of 2% or somewhat higher close to the cloud centerline. An additional 8 patients resided in the area but were mapped at their workplaces. If some of them were actually at home when exposed, the average attack rate for the area could have been about 3%.
           The automotive maintenance and repair center, located about 3.9 km directly downwind of the source, had three anthrax patients (Figure 2, patients 45, 46, and 62). From a cursory tour of the center, I estimate that it had 200-300 employees. The corresponding attack rate would be about 1%, subject to errors of sampling and ascertainment.
           It appears from the above that the attack rate of 1-2% recorded at the ceramics factory is consistent with the rates estimated for locations upwind and downwind of it, providing some assurance that the ceramics factory rate used in the article for source strength estimation is not anomalous.
            Dose-response relation. The largest uncertainty in estimating the source strength is not in the model of atmospheric dispersion or in the estimate of the anthrax attack rate but in the relation between the number of viable spores inhaled and the probability of contracting inhalation anthrax. Even if there were an agreed dose-response relation for non-human primates, which there is not, there would remain uncertainty regarding the relation applicable to the actual population at risk and the particular aerosol encountered at Sverdlovsk.
            In order to reflect the uncertainty, the article employs two different dose-response relations. These are described in the text and in Table III:

1. Log-normal with a human LD₅₀ of 8,000 spores (8) and a slope of 0.7 probits per log dose (26,35). The LD₅₀ comes from a U.S. Department of Defense estimate that the human LD₅₀ is 8000-10,000 spores (8). The probit slope is that reported by Glassman (26, 35) for a large-scale experiment with cynomolgus monkeys.
2. Independent spore action (38) with LD₅₀ = 45,000. The LD₅₀ is that reported by Druett et al. for rhesus monkeys (37).

           A log-normal relation allows for heterogeneity in susceptibility among individuals in the exposed population. This is particularly important at low attack rates, as at Sverdlovsk, since the more susceptible members of the population will be over-represented among those who contract anthrax.
           The independent action relation is based on the reasonable assumption that spores act independently, not cooperatively, in the initiation of inhalation anthrax. But the model assumes that there is no heterogeneity of susceptibility in the population. At low attack rates, the neglect of heterogeneity will cause the dose to be overestimated. The estimate from the independent action dose-response relation, with a high LD₅₀ (45,000 spores) and no allowance for heterogeneity, is included in the Science article in order to present an extreme high-end estimate.
           Source strength. Table IV gives source strengths, in milligrams, calculated from three different dose-response relations and two different atmospheric dispersion models for an attack rate of 2% at a distance of 2.8 km directly downwind of the source and a breathing rate of 30 L/min.
           In addition to the two dose-response relations used in the article, the table gives source strengths estimated on the basis of independent spore action with LD₅₀ = 8,000. The doses for a 2% attack rate, calculated dose-response relations A, B, and C, are 9.3, 230, and 1300 spores, respectively. The number of spores per milligram is taken as 10⁹ (39).
           The two estimates given in the Table for dose-response relations A and C with the Briggs "D" atmospheric dispersion model are those given in the Science article.

Matthew Meselson

Editor's Note: Prof. Meselson has been at Harvard long enough to be classified a true New Englander and this evening his true New England spirit was tested to the max. ASA cajoled the Professor into going over this paper until almost 0200 hours on 12/17 (we had to be at the printer at 0800). Matt - many thanks.

References:
Note: All numbered footnotes refer to the article listed below.

The Sverdlovsk Anthrax Outbreak of 1979
Matthew Meselson, Jeanne Guillemin, Martin Hugh-Jones, Alexander Langmuir, Ilona Popova, Alexis Shelokov, Olga Yampolskaya

Science, New Series, Vol. 266, No. 5188. (Nov. 18, 1994), pp. 1202-1208.
Stable URL: http://links.jstor.org/sici?sici=0036-8075%2819941118%293%3A266%3A5188%3C1202%3ATSAOO1%3E2.0.CO%3B2-1

TABLE I
BASIS OF DOSE CALCULATIONS

Atmospheric stability "D"
Wind speed circa = u = 5 m/sec at 10 m
Release height = 10 m
Source strength = Q spores
Deposition negligible
Infectivity independent of travel time
No mixing layer
Breathing rate = B = 5 X 10^-4 m³/sec (= 30 L/min)

Dose = [QB][pus_ys_z]-1exp[-(1/2)(y/s_y)²]exp[-(1/2)(10/s_z)²]
p = [3.18 X 10^-5Q][s_ys_z]-1exp[-(1/2)(y/s_y)²]exp[-(1/2)(10/s_z)²]
s_y = [0.08x][1+0.0001x]^-1/2
s_z = [0.06x][1+0.0015x]^-1/2

Downwind (x) and crosswind (y) distances are in meters.

TABLE II
CENTERLINE DOSE (SPORES) FROM FOUR MODELS

Wind speed circa = 5 m/sec
Breathing rate = 30 L/min
Source strength = 10¹⁰ Spores

note: Stability "D" is neutral. Stability "E" is slightly stable

TABLE III
DOSE-RESPONSE RELATIONS