Why the Independent Variable in Virology Remains Unestablished

Paid subscribers get one week of early access to all Deep Dive articles and AntiViral episodes, along with several other subscriber benefits. A full-year subscription is currently 25% off. [See everything included with a subscription here.]

Virology rests on a foundational methodological flaw, one that has persisted since the very birth of the field, and it strikes at the core of scientific inquiry itself: how causation is established. The flaw is so severe that it erodes the very ground upon which virology stands. Exposing it completely dismantles the claims that sustain the discipline. This is why the question of what it takes to scientifically demonstrate causation has remained central to my critique of virology.

However, before examining virology’s reliance on this error, we must first clarify what it actually means to demonstrate that something causes a specific outcome. Not merely suggest it. Not simply correlate with it. But prove it as a true cause‑and‑effect relationship. That question lies at the heart of what makes a hypothesis scientific, and understanding this process makes virology’s fatal flaw impossible to overlook.

The Anatomy of a Hypothesis

Evaluating any claim of cause and effect requires a clear definition and understanding of what constitutes a scientific hypothesis. As defined by Britannica, a scientific hypothesis is a tentative explanation about a phenomenon or a narrow set of phenomena observed in the natural world. At its core, it proposes that a specific, identifiable factor produces a particular outcome. But testing that proposal requires one essential step: the cause must be isolated and manipulated on its own, clearly distinct from alternative explanations.

To see how this works in practice, a scientific hypothesis begins with an observed natural phenomenon—an effect or outcome that occurs in the natural world and requires explanation. This observed outcome is known as the dependent variable (DV), the phenomenon the researcher seeks to understand. For example, if a tree catches fire, the fire itself represents the observed outcome (DV) the researcher wants to explain.

The next step is to identify a proposed cause, known as the independent variable (IV). In this example, the hypothesis might propose that lightning striking the tree caused the fire. For this proposed cause to be scientifically tested, it must be identifiable, present, and capable of being independently controlled or manipulated to determine whether it can actually produce the observed effect.

In addition, all other possible causes, known as confounding variables (CV), must be accounted for. In this example, that could include a lit match, a discarded cigarette, or even environmental factors like heat buildup from dry leaves.

Only by isolating the proposed cause and controlling for alternative explanations can the researcher determine whether the effect is truly the result of that cause, rather than an assumption based on correlation or incomplete evidence.

This principle is reflected in the 2018 paper From Ideal to Real Risk: Philosophy of Causation Meets Risk Analysis. In section 4.2. Causation Must Be Studied by Comparison and in Isolation, it is emphasized that causal factors must be isolated or controlled in order to determine their specific contribution to an outcome. The authors write:

“In the lab, one will try to isolate the causal factor as much as possible from its usual complexity to see what it does on its own, but this might not always be practically possible. RCTs are one way to do this, the idea being that if there is a difference in outcome between the two situations, it is because of the controlled factor. Only if the cause can be isolated or controlled in this way, can we say to have “observed” its causal contribution. Without such isolation, we cannot know if some other factor was present that produced the effect instead. This is called the problem of overdetermination and is a problem for studying causation in its natural complexity (Collins, Hall, & Paul, 2004). From an empiricist point of view, we thus see that causal complexity becomes a challenge for establishing what exactly the causal contribution of each factor is, which means that separation and isolation are necessary steps of the scientific procedure.”

The authors further acknowledge that individual factors must be distinguished from confounders if we are to determine whether they actually contribute to the outcome. This is one of the basic difficulties of establishing causation. Natural systems are complex, and unless the variable being examined can be isolated, we cannot simply assume that the observed outcome was caused by that variable rather than something else.

This provides a straightforward standard for determining whether causation has actually been demonstrated. Observing an outcome and associating it with a proposed cause is not enough. The cause must be identified and isolated, and shown to produce the outcome without relying on other possible explanations.

The Inversion of Logic

It is upon applying this scientific standard to “infectious” disease research that an important question is raised: does virology follow the basic rules of experimental science? Specifically, do researchers isolate and test the independent variable before observing its alleged effects? This is where the fatal methodological flaws arise in virology.

​While there is an observed phenomenon—illness and its apparent “spread”—and a clearly definable dependent variable in the form of symptoms, the corresponding independent variable is not established in a direct and controlled manner. Examining the historical literature reveals that the concept of a “virus” did not arise as a discovered entity, but as a theoretical construct, a late 19th century placeholder explanation for disease processes that bacteria could not explain.

​Unlike a lightning strike, which can be directly observed and tested independently of the fire it causes, the “virus” was inferred from the very effects it was proposed to explain. Because the independent variable was not independently established and characterized apart from the effect it was claimed to produce, the foundational requirement for a true cause-and-effect demonstration was bypassed from the very start.

Instead, the effect—disease or cellular changes—is reproduced under artificial laboratory conditions using complex, unpurified materials such as fluids, tissues, or swabs. These materials are, by definition, a collection of confounding variables containing countless proteins, extracellular vesicles, cellular debris, and unknown factors. When these are introduced into animals or cell cultures through unnatural routes of exposure, any observed effect cannot reliably be traced to a single, isolated variable. The cause is inferred from the effect rather than demonstrated as a distinct and manipulable variable capable of producing the effect on its own.

The requirement for a purified agent to prove causation is not some arbitrary requirement—it dates back to the mid-19th century. Yet throughout the birth of the field, researchers routinely acknowledged their inability to achieve full purification free of confounding factors.

​While defenders of virology argue that strict physical isolation is unnecessary or unreflective of biological research standards, evaluating that claim requires examining the record itself: the standards established at the birth of the field, the admissions made across decades of literature, and where the methodology stands today.

You may be surprised to learn that, despite unprecedented technological advances, virology still operates without this critical missing variable required to truly demonstrate causation. The fatal flaw remains unresolved.

What follows is an intensive examination of this problem, broken into six parts within this article. Each part builds upon the last, tracing the historical development of the methodology, the attempts to establish the “virus” as an independent causal agent, and the problems that remain unsettled today. It is a lengthy investigation, but one that requires time to fully examine and digest the evidence.

By the end, you will be asking: Where is the missing variable?

1. The Purity Standard

“It then remains to prove that the isolated parasite is really the cause of the disease. To accomplish this, one must show that animals inoculated with the pure culture contract the original disease. The inoculation must succeed not only sometimes but in every attempt…”

—Robert Koch, Ueber die Aetiologie der Tuberkulose, 1882

The Henle-Klebs Mandate

Decades before Koch made this statement, in the 1840 work Pathologische Untersuchungen, his teacher Jakob Henle emphasized the necessity of isolating a causal agent in order to study its effects independently. He wrote that “one could prove empirically that [the organisms] were really effective only if one could isolate the contagious organisms from the contagious fluids, and then observe the powers of each separately.” In other words, to ascribe causation to a specific microorganism, it must first be separated from all other potential causes of the same effect.

This principle was echoed in 1872 by the German-Swiss microbiologist Edwin Klebs, who noted that simply “tracing the invasion and the course of the micro-organisms can make causality probable, but the crucial experiment is to isolate the efficient cause and allow it to operate on the organism.”

Together, Henle and Klebs established an important methodological principle that would become central to microbiology: the proposed cause (IV) must be separated from confounding factors (CV) so that its effects (DV) can be tested on its own.

Koch’s Early Concessions

However, this standard proved difficult to meet in practice during the early development of bacteriology. Throughout his work on anthrax in the mid-1870s, Robert Koch did not have a definitive method for ensuring complete purity of his cultures. In his paper Robert Koch: From Obscurity to Glory to Fiasco, microbiologist Martin Dworkin pointed out that Koch’s anthrax work preceded the development of his plating technique for obtaining isolated colonies and was therefore conducted without the use of isolated pure cultures. As historian K. Codell Carter noted in The Koch–Pasteur Dispute on Establishing the Cause of Anthrax, Koch’s reliance on microscopic observation to rule out contamination would not have satisfied many of his contemporaries, who regarded precise isolation techniques as essential.

In fact, Koch himself acknowledged the difficulty of meeting this standard. In his 1881 paper Zur Ätiologie des Milzbrandes, he described the demand to isolate bacilli from all associated substances as effectively “impossible,” remarking that “no one can take seriously such an undertaking.” He further conceded that, because of the lack of purity, “one could still object that … some other substance associated with the bacteria, rather than the bacilli themselves, causes the disease.” Koch ultimately concluded that, for all practical purposes, the purity objection to his anthrax work was “meaningless.”

His contemporaries, however, did not share this view. As noted by Carter, critics of Koch’s work regarded this objection as fundamental. In reviewing Koch’s paper, Professor Zuber observed:

“How opinions can differ! We think, and much of the world with us, that this objection has a capital importance, and we reproach precisely those authors who…subject complex liquids to complicated procedures and thereby reach dubious results.”

Zuber went on to praise those who were attempting to completely isolate the bacillus through successive cultures, emphasizing the importance of achieving true separation from all other substances.

It was clear that Koch’s contemporaries understood the importance of isolating a single causal factor before attempting to establish causation. The presumed cause had to be separated from other influences so that its effects could be assessed independently. This was precisely where Koch faced criticism: he had not fully demonstrated that the anthrax bacillus alone, independent of the other material involved in his experiments, was responsible for producing the disease. For all intents and purposes, he had not yet conclusively established a causal relationship, and his evidence was regarded by many as insufficient.

The Methodological Pivot of 1881

Against this backdrop, Koch’s 1881 publication Zur Untersuchung von pathogenen Organismen marked a major shift in methodology. Having previously dismissed complete isolation as practically impossible, he now argued that isolating a single organism from all others was essential for proper study. He wrote:

“And once it has been established that a notoriously harmful or even merely suspicious organism is present, the next task is to study it separately from others—which would otherwise interfere with and confuse observation—in all its characteristics: to investigate its conditions of life, its developmental history, and everything that promotes or inhibits it.

However, such knowledge can only be obtained with the help of continuous cultures of the individual species, so-called pure cultures, for which, up to now, no methods exist that are universally applicable and reliable in every respect.”

Here, Koch directly identified the problem of confounding variables (“others which would interfere”) and the necessity of isolating the proposed causal factor for testing.

To address this, Koch developed improved techniques for obtaining pure cultures using solid media (initially gelatin) on glass plates, allowing individual colonies of bacteria to be separated and propagated. This process involves taking a mixed sample, distributing it across a solid surface to isolate individual colonies, and repeatedly subculturing a single colony to ensure purity. Koch later applied these methods in his work on tuberculosis, where he reported obtaining the tubercle bacillus in pure culture, consisting of a single species without admixture.

In his 1882 work Ueber die Milzbrandimpfung, Koch further emphasized that demonstrating causality requires freeing the organism from all diseased material before testing it:

“Only after one has, in this way, gained a thorough orientation as to whether microorganisms are present in the diseased parts, and at which locations they are found in complete purity—for example in the lung, spleen, heart blood, etc.—can one attempt to prove that these microorganisms are pathogenic in nature and that they are specifically the cause of the disease in question.

For this purpose they must be cultured pure and, after they have thereby been entirely freed from all the parts of the diseased body, they must be inoculated back into animals, preferably of the same species as those in which the disease was originally observed, or at least into such animals in which the disease in question is known by experience to occur with unmistakable symptoms.”

The Requirement of “Perfect Proof”

By his 1884 paper Die Aetiologie der Tuberculose, Koch went even further, stating that the only way to obtain “perfect proof” of causation was to completely separate the organism not only from the diseased host but also from all potentially confounding disease products:

“The facts obtained in this study may possibly be sufficient proof of the causal relationship, such that only the most sceptical can raise the objection that the discovered microorganism is not the cause but only an accompaniment of the disease. However, many times this objection has a certain validity, and then it is necessary to obtain a perfect proof to satisfy oneself that the parasite and the disease are not only correlated, but actually causally related, and that the parasite is the actual direct cause of the disease. This can only be done by completely separating the parasite from the diseased organism, and from all of the products of the disease which could be ascribed to a disease-inducing influence, and then introducing the isolated parasite into healthy organisms and inducing the disease anew with all its characteristic symptoms and properties.”

Koch clearly distinguished between mere correlation and true causation, insisting that only complete separation of the organism from all other disease-associated material could establish a direct causal relationship.

Koch’s later emphasis on pure culture methods reinforced the need to isolate the organism from other potentially confounding material. He recognized that partial or uncertain separation would leave open the very objection that had been raised by his critics, that some other associated substance, rather than the organism itself, might be responsible for the observed effects.

In his attempts to address his critics, Robert Koch would later emphasize the central importance of pure culture techniques, writing that “the pure culture is the foundation for all research on infectious diseases.” This principle remains foundational in modern microbiology. As the Journal of Visualized Experiments (JoVE) explains:

“Pure cultures, defined as the growth of a single microorganism species isolated from mixed populations, are fundamental tools in microbiological research…allowing researchers to study microbial traits under controlled conditions.”

Early bacteriology therefore treated isolation as a methodological requirement: the proposed causal agent had to be separated from a mixed population before its role in disease could be evaluated. As microbiologist Martin Dworkin observed, Koch’s introduction of solid media for bacterial cultivation “allowed the emergence of pure culture microbiology” and, together with the later canonization of Koch’s Postulates, “reified an attitude about the microbial causation of disease.” While achieving this in practice could be challenging, the underlying principle was clear. However, even as these advances were codified, the question of whether such isolation could be consistently achieved would soon be tested by the emergence of the concept of “filterable infectious agents,” where the definition of isolation would be forced to change entirely.

2. From Purity to Inference

“Purity means the degree of freedom from nonviral components, or, conversely, the extent to which viral particles show gross physicochemical homogeneity.”

— Chemistry of Viruses

The Birth of a Placeholder

The birth of virology began by breaking from bacteriology’s standard of purity. The concept of a “virus” took shape in the late nineteenth century through the work of Dmitri Ivanovsky and Martinus Beijerinck on tobacco mosaic disease in plants. Both researchers were unable to associate the disease with observable bacteria. However, they reported that disease transmission could still occur when plant sap was passed through filters designed to retain bacteria and then introduced into healthy plants through mechanical injury.

As noted by Fields Virology, it was only when bacteriologists failed to prove bacteria as causative agents of disease using Koch’s Postulates that the concept of a “virus” was born. Unlike bacteria, which could be directly observed under a light microscope, these proposed agents were not visible and could not be independently isolated. Instead, their presence was inferred from the effects observed after introducing filtered material into living organisms.

This reliance on inference rather than direct observation was not lost on early bacteriologists. In the inaugural address at the Academy of Sciences on July 1, 1909, Robert Koch himself acknowledged the existence of “infectious” agents that could not be visualized or isolated using available methods:

“However, there are microorganisms that are far smaller still. They pass easily through fine-pored filters that retain all bacteria…We must infer their existence because they cause various diseases…”

Koch noted that these agents could not be seen even with the best microscopes of the time and that their nature remained unknown.

“At present, we cannot form any clear idea…what kind of organisms they are…”

​By acknowledging this, the founder of modern bacteriology fundamentally changed the rules: scientists were no longer confirming the alleged “pathogen’s” presence before the experiment, but inferring it from the result afterward. The independent variable had moved from the realm of the observable to the realm of the presumed.

The Redefinition of Purity

The inability to see or isolate these agents necessitated a radical departure from the existing scientific method. In A History of Experimental Virology, Alfred Grafe acknowledged that because these agents could not be grown on synthetic media, the foundational requirement of isolation became an impossibility:

“Since filterable viruses could not be grown on synthetic media, in vivo and in vitro experiments with pure cultures were not possible, and thus the isolation of a single pathogen was virtually impossible.”

​Grafe noted that because these agents were “readily mixing with other germs in the host,” the prospect of growing them in a pure state was considered “grasping for the stars.” This technical impasse led to a “bevy of potential errors,” as presumed microbes were classified as “viruses” simply because it was assumed that they passed through a filter and appeared to transmit “infection.” He admitted that “far more pathogens were classified among the filterable viruses than could convincingly be proved as such.” In other words, the label “virus” became a container for everything that could not be “proven” through traditional isolation. This impasse forced a redefinition of “purity” itself. While a pure culture in bacteriology referred to a single organism isolated on a plate, Grafe revealed that in virology:

“By a pure suspension of filterable virus is meant a statistical population responding uniformly to filtration or to centrifugation.”

​This inverted the standard experimental sequence. Rather than isolating an independent variable from the outset, researchers relied on crude samples where the alleged cause was inferred entirely from the observed effect.

Rivers’ Early Warnings

These concerns regarding the limitations of the filtration methods used to infer the existence of these entities were echoed by eminent virologist Thomas Rivers. Writing in his seminal 1927 paper Filterable Viruses: A Critical Review, Rivers acknowledged the imprecise nature of the methods:

“Methods of filtration are crude and inaccurate and the most one can say regarding the viruses is that under given experimental conditions they either pass or do not pass through certain filters.”

He noted that numerous factors—such as protein content, cell debris, electrical charge, and experimental conditions—could influence whether material passed through a filter, making it difficult to draw clear conclusions about the nature of the agent involved:

“Sometimes, however, small bacteria may still contaminate the filtrates or two viruses may be present in the filtered material. Therefore, filtrates from uncontrolled and even from well controlled sources may contain more than one active agent, some of which may be cultivated on simple or on special media.”

Rivers cautioned investigators against over-interpreting such results​:

“All investigators should be extremely careful in working with filtrates not to be misled by their findings and ascribe to an active agent an etiological role in a disease with which it has nothing more than-accidental connection.”

He was essentially admitting that, with the available methods, the presumed independent variable was inseparable from the known confounding variables. He warned that researchers might easily ascribe a “causal role” to something that had nothing more than an “accidental connection” to the disease.

Rivers further addressed the issue of the purity standard in “viral” preparations, noting no “virus” had ever been obtained in an “absolutely pure state.” He emphasized that it was therefore impossible to determine whether filtration separates a single agent or a mixture of material, including protein aggregates and degraded cellular debris:

“No virus has been obtained in an absolutely pure state. Not even the washed granules of vaccine virus can be accepted as representing only virus. Therefore it is impossible to say that virus alone is being filtered rather than virus attached to aggregates of protein or particles of degraded cells.”

Rivers also noted that the physical nature of “viruses” had not been satisfactorily resolved, and attributed this uncertainty to the fact that the particulate material observed in experiments, whether obtained through filtration or centrifugation, could not be assumed to represent a single, isolated entity:

“The question of the organized or corpuscular nature of the viruses has not been satisfactorily settled. This is due to the fact that most workers realize that the granules which are seen, which are frequently held back by filters, and which are thrown down by prolonged centrifugation may not represent virus alone. This is particularly true since viruses exhibit a remarkable tendency to be adsorbed by many things with which they come in contact.”

Rivers’ early warnings highlighted a huge practical problem: so-called “viral” preparations could not be reduced to a single isolated entity as was standard in bacteriology. The samples contained multiple biological components that could not be fully separated with the available technology of the era.

The persistent failure to achieve purification meant that every physical property attributed to these agents, from their size to their electrical charge, remained speculative. This was candidly addressed by eminent bacteriologist and physician Dr. Earl McKinley in a 1932 article in the journal Science, which noted that while filtration experiments were “exceedingly worth while” for general knowledge, they had failed to reveal the actual nature of the agents:

“What is known to-day as a result of studies on cataphoresis to determine electric charge can not be regarded as conclusive for the reason that, up to the present time, no virus has been obtained in a pure state. Even the size of viruses is at present only an approximation.”

​McKinley admitted that the field was operating not on proven entities, but on a “working hypothesis”—the presumption that a “virus” is a particulate, animate agent simply because an agent is assumed to pass through ordinary filter candles. The article concluded with a revealing admission: even if the actual size were known, it would not “aid us materially” in understanding the nature or mode of action of these agents compared to known entities like bacteria.

That same year, Thomas Rivers continued to address the ongoing challenges in the study of so-called “viruses” in his paper The Nature of Viruses, noting that experimental preparations were often dominated by non-specific biological material. He pointed out that physical measurements made on such preparations, such as electrical charge, may not reflect the properties of the presumed “infectious agent” itself, but rather those of associated material acting as carriers. He expressed skepticism regarding whether true separation had been achieved:

“Others, however, aware of the fact that virus preparations usually consist principally of proteins and bits of degraded cells from the host, realize that the electrical charges determined might not be those of the virus particles themselves but of their carriers, i.e., material on which the virus particles are adsorbed. It is true that a few experiments have been performed with “protein-free” preparations of viruses. But an examination of the methods of purification fails to convince one that such purified viruses had been completely separated from their carriers.”

Rivers emphasized that “viral” preparations were recognized as complex mixtures of non-specific biological material, which made purification a necessary methodological objective. He therefore argued that obtaining “viruses” in a pure or relatively pure state was a natural and necessary aim of the field. He stressed that such purification was not merely desirable but essential for meaningful investigation of key properties such as size, electrical charge, and immunological response:

“Inasmuch as virus-containing emulsions consist chiefly of substances unrelated to the active agents themselves, it is natural that workers should attempt to obtain the viruses either in a pure or in a relatively pure state. Moreover, it is being realized that purified viruses are essential for the proper study of problems in this field, such as the estimatlion of the size of viruses, the determination of their electrical charge, and the investigation of immunological responses excited by them.”

He further noted that many purification techniques were adapted from chemical and enzymatic work, yet still had not reliably produced fully purified “viral” material:

“Most methods of purification of viruses are based on the principles of precipitation by a variety of chemicals and selective adsorption and elution as used extensively in enzyme work. As yet, it is unlikely that a virus has been obtained in a state of absolute purity. Nevertheless, the results already secured are encouraging and should excite further investigations. It may be possible in this way to attain eventually a more accurate concept of the nature of some viruses.”

Rivers pointed out a critical problem: without a pure sample, researchers could not answer basic questions. He presented a field still struggling with the distinction between presumed “infectious agents” and the complex biological materials in which they were supposedly embedded. His renarks highlighted the urgent need for better separation techniques.

The Relaxing of the Rules

However, a few years later in 1937, Thomas Rivers published the paper Viruses and Koch’s Postulates, where he made a crucial admission: “It is obvious that Koch’s postulates have not been satisfied in viral diseases.” In addressing the role of purification and its relationship to Koch’s framework, he noted that the requirement to cultivate an “infectious agent” in a pure state on lifeless media was holding the field back:

“The idea that an infectious agent must be cultivated in a pure state on lifeless media before it can be accepted as the proved cause of a disease has also hindered the investigations of certain maladies, inasmuch as it denies the existence of obligate parasitisms the most striking phenomenon of some infections, particularly those caused by viruses. Moreover, it ignores the possibility that certain viruses may be fabricated autocatalytically in living cells.”

Turning his back on his previous statements about the necessity of obtaining a “virus” in a pure state, Rivers then suggested that the requirement to isolate an agent in a pure state should be relaxed for “viruses.” He proposed using modified tissue cultures as a substitute for pure culture:

“One might say that the present-day method of propagating viruses in modified tissue cultures should be considered as taking the place of cultivation on lifeless media.”

However, he immediately undermined his own suggestion:

“I doubt whether the substitution is warranted, because the principles underlying the two methods of cultivation are radically different and Koch certainly did not have tissue-culture methods in mind when he proposed his rules of proof.”

While Thomas Rivers reconsidered the classical requirement of pure culture in the context of virology, his proposed alternative did not operate on the same rigorous experimental principles defined by Koch. In doing so, he relaxed several elements of the original criteria, noting that it was not obligatory to demonstrate the presence of a “virus” in every case of the disease, while also recognizing the existence of asymptomatic (i.e., healthy) carriers. He further stated that it was not essential for a “virus” to be grown on lifeless media or even in modified tissue cultures. By relaxing these criteria, Rivers allowed the field to bypass the need for a purified IV.

This clearly marks a departure from his earlier recognition that purified material was essential for properly investigating the properties of these agents. Although Rivers maintained that experimental material “should be free from ordinary microbes,” proposing filtration as a means of achieving this, he had already acknowledged that such methods were crude and incapable of ensuring that a single, isolated agent was being studied. As a result, the original standard was not replaced with an equivalent one, leaving unresolved the question of whether a fully controlled and independently established causal variable had ever been achieved in virology.

Calculating Purity: The 1:1000 Ratio

Two years after acknowledging that these essential standards had not been fulfilled, the problem of purity remained unsolved. In the 1939 paper Estimation of the Purity of Preparations of Elementary Bodies of Vaccinia co-authored by Thomas Rivers, the authors openly admitted:

“Investigations on the nature of viruses were hampered for many years by the impurities of virus suspensions.”

At this stage, the discussion shifted toward developing indirect methods for estimating “purity.” The authors proposed that purity could be assessed by comparing the number of “infective units” to the total number of observed particles:

“Knowledge in this field has advanced to the stage where we believe that such an estimation can be made by the determination of the ratio of the number of infective units in a given preparation to the total number of the virus particles…”

When this method was applied, the results themselves highlighted the severity of the problem. The authors reported:

“If one uses the values 10 cc., 10⁸ units in 0.2 cc., and 25 mg. to represent the volume, titer, and dry weight of a lot of virus and employs these in a calculation of the type applied to the present experiments, one obtains a ratio of the order of 1:1000. Such a ratio, we believe, indicates the presence of an excessive amount of impurities in the preparations. Moreover, it is obvious that such preparations should not be used for studies of the nature of the virus of vaccinia.”

By their own methodology, “purity” was not established through direct isolation of a single, defined entity, but inferred through indirect calculations. The estimated number of “elementary bodies” was derived from the dry weight of the preparation, based on an assumed weight for a single particle and the assumption that the material consisted entirely of such bodies.

Even when relatively close ratios between “infective units” and total particles were obtained, this did not independently verify the identity or purity of the particles themselves. It reflected a calculation that presupposed the identity of the particles being measured.

​The authors acknowledged that extreme particle ratios—around 1:1000—pointed to heavy contamination, making those samples useless for studying the true nature of the “virus.” Because researchers could not actually separate the “pathogen” from host tissue, they simply worked around the problem. Purity was no longer directly proved; it was deduced by comparing total particle counts against observed “infection” rates.

3. The Pursuit of Purity

A Fundamental Failure of Control

Following the concessions made by Thomas Rivers, the virology literature of the 1920s through the 1940s became a record of failed attempts to separate the “agent” from the “host.” For example, in early studies on bacteriophages, the “viruses” said to “infect” bacteria, researchers acknowledged that standard experimental preparations consisted of complex mixtures.

In the 1929 paper The Preparation of Relatively Pure Bacteriophage, the authors noted that phage preparations contained the proposed agents along with nutrient constituents of the broth and substances released from lysed bacteria. They emphasized that such impure preparations were not suitable for studying the physical and chemical properties of bacteriophage, arguing that it was necessary to use purified preparations or, alternatively, to account for the influence of the surrounding medium:

“Studies on the nature of the bacteriophage are usually carried out with lysates which contain, in addition to the lyric principle itself, the nutrient constituents of the broth and substances set free from the lysed bacteria. While such preparations are suitable for many types of investigation, there are certain problems, e.g., those dealing with the physical or chemical properties of the bacteriophage, which require the use of relatively pure phage. That the composition of the medium in which bacteriophage is suspended exerts considerable influence on the physical state of the lyric agent was brought out by recent work of the authors (1), who showed that purified bacteriophage free from bacterial proteins not only has a smaller particle size than non-purified phage but also will readily unite with proteins or protein derivatives to form the larger particles found in ordinary lysates. This point, that the lyric agent is commonly associated with colloidal aggregates has been repeatedly stressed by Bronfenbrenner (2) and it appears quite logical that physical-chemical experiments dealing with the phage itself should either be performed with purified bacteriophage free from the influence of a complex suspending medium such as nutrient broth, or else that the influence of the menstruum must be taken into account.”

The authors also noted that studies examining therapeutic effects required purified preparations in order to avoid non-specific reactions due to other constituents of the lysates:

“It is suggested that experiments designed to test the therapeutic value of bacteriophage be conducted, when possible, with purified suspensions thereby avoiding any possibility of obscure non-specific reactions due to other constituents of the lysates.”

This was not just a minor technical flaw—it was a complete breakdown of basic scientific controls. When preparations remain heterogeneous mixtures, it is logically impossible to verify that a single agent (IV), rather than the complex mixture of host debris (CV), is responsible for the observed effect (DV).

Sabin’s Search for Mode of Action

These observations were shared by Albert Sabin (known for his oral polio vaccine) in his 1932 paper Experiments on the Purification and Concentration of the Virus of Poliomyelitis. Sabin noted that separating the “virus” from tissue substances was necessary in order to gain a thorough understanding of its mode of action and nature, while acknowledging that the available methods were only effective in removing most associated tissue constituents:

“For a thorough understanding of the nature and mode of action of filterable viruses, it is necessary that they be separated from the tissue substances with which they are associated. Moreover, the methods whereby such purification may be accomplished, are in themselves capable of fielding additional data concerning the properties of the virus in question. The purpose of this communication is to describe procedures effective in separating the virus of poliomyelitis from most of its associated tissue constituents and in its subsequent concentration.”

Sabin also acknowledged that immunological and chemical studies were either impossible or unreliable using the concentrations obtained by ordinary extraction methods:

“Certain immunological and chemical studies on filterable viruses are either impossible or unreliable with the concentrations obtained by the ordinary methods of extraction.”

The Sedimentation Trap

In the 1942 paper Studies on the Purification of Poliomyelitis Virus, virologist Hubert Loring admitted that, despite high-speed centrifugation, no evidence had been supplied showing that normal host materials were not sedimented along with the “viral” particles. In fact, based on the work of other investigators, Loring suggested that this contamination was likely the case:

“While it has been shown that high speed centrifugation resulted in a sedimentation of the virus, evidence has not been provided that other macromolecular materials associated with normal tissue were not also sedimented along with the virus under these conditions. In view of the work of several investigators (4–6) on the concentration of macromolecular materials from various normal animal tissues, such constituents might very well be present together with virus in the high molecular weight fractions obtained from poliomyelitis-infected tissues.”

He noted that purification of the “virus” required that the extracts were free from tissue fragments and lipoidal material, as well as both low- and high-molecular-weight soluble constituents:

“Purification of the virus by differential centrifugation requires (1) the preparation of large quantities of infectious extract free from tissue fragments and lipoidal material, (2) the separation of the virus from soluble constituents of low molecular weight, and (3) the separation of the virus from high molecular weight substances which are thrown down with the virus in the ultracentrifuge.”

These limitations closely followed the earlier concerns expressed by Albert Sabin regarding the difficulty of obtaining sufficiently purified material for meaningful biological and chemical study. Despite advances in separation techniques such as centrifugation, the problem of distinguishing the presumed “virus” from host-derived macromolecular substances remained unsettled, with no clear methodological assurance that true isolation had been achieved.

Stanley and the Size Discrepancy

In the 1943 paper An Evaluation of Methods for the Concentration and Purification of Influenza Virus, virologist Wendell Meredith Stanley, who is credited with “purifying” the tobacco mosaic “virus” through crystallization in 1935, similarly emphasized the importance of purification for studying the fundamental physicochemical properties attributed to “influenza virus” preparations. He noted that uncertainty regarding basic properties of the agent reflected the lack of adequately purified material available for study, particularly in relation to its estimated size. Stanley wrote:

“The concentration and purification of influenza virus are of special importance in connection with studies on the fundamental chemical and physical properties of the virus itself. The lack of knowledge of such properties is emphasized at the present time by the disagreement concerning so simple and yet so fundamental a property as the true size of the infectious unit of influenza virus. For some years it was generally accepted that the virus was represented by a sphere about 80 to 120 mμ in diameter (1, 2). Recently, however, Chambers and coworkers (3) and Bourdillon (4) secured data, some of which reasonably could be interpreted as indicating that at least a portion of the virus possesses a size of the order of 10 mμ. Friedewald and Pickels (5) and Taylor and associates (6), on the other hand, consider, on the basis of recent experiments, that the size of the virus cannot be 10 mμ and is more nearly that accepted earlier.

He concluded that purification was necessary in order to enable experimental correlation between a physical entity and biological activity:

This general situation, which is illustrative of the lack of definite knowledge of the fundamental physicochemical properties of influenza virus, emphasizes the necessity for studies on the concentration and purification of the virus, for with the preparation of purified virus it should become possible to conduct experiments in which a given physical entity is correlated with virus activity. The direct positive correlation of virus activity with a given physical entity on fractionation of that entity by different procedures would provide good evidence that the preparation is essentially pure and hence suitable for studies on its fundamental physicochemical properties.”

Stanley also noted the practical importance of purification in the context of vaccine production using chick embryo material, emphasizing the need to reduce contaminating host-derived components:

“The widespread use of the chick embryo for the cultivation of viruses (21) and the current increase in the use of infectious materials from this source for human vaccination make it prudent to attempt to purify influenza virus obtained from chick embryos in order to avoid or reduce difficulties due to sensitization of human beings to chick embryo proteins.”

Differential centrifugation was presented as a preferred method for achieving such preparations, yielding material “presumably free of large amounts of inactive protein” and suitable for both experimental study and large-scale applications:

“Differential centrifugation provides a direct means of obtaining a concentrated and purified preparation, presumably free of large amounts of inactive protein, and therefore appears to represent the method of choice for the large scale preparation of purified virus for purposes of vaccination or for the study of physicochemical properties.”

Stanley’s observations fit squarely within a broader pattern emerging in the virology literature at that time: purification was not treated as a peripheral concern, but as a necessary condition for resolving even the most basic questions about the nature of the agent under study. His emphasis on correlating a specific physical entity with “virus” activity reflects the need to establish a defined independent variable, one sufficiently separated from surrounding biological material to be tested on its own. At the same time, his cautious language, particularly his reference to “presumably” purified preparations, shows how difficult it remained to achieve that standard in practice.

The Golden Jubilee Admissions

In the inaugural paper for the 1950 Symposium on Viral and Rickettsial Diseases held at the Golden Jubilee meeting of the Society of American Bacteriologists, Joseph E. Smadel acknowledged that “viruses” remained poorly understood and that experimental preparations were dominated by host-derived material. He emphasized that this lack of purification posed a fundamental obstacle to studying their properties:

“The nature of viruses remains an enigma but curiosity continually stimulates investigators to attempt to understand these most minute of living things. Viruses and rickettsiae will grow only in living cells, hence, all initial preparations of these agents are heavily contaminated with normal cellular components. Indeed, it would be more accurate to say that preparations of host cell material are contaminated with virus. The need for purified preparations of viruses for studies on the physical, chemical, biological and immunological nature of these substances is apparent. Hence, it is not surprising that many investigators have devoted much time and effort in attempting to attain these. Dr. Warren discusses the results which have been obtained in the purification of the animal viruses, in general a more recalcitrant group for such studies than the plant viruses. Furthermore, he devotes considerable time to the newer methods which may be useful in eliminating normal components from preparations of the small, mammalian viruses, a group which has been particularly difficult to purify.”

Smadel essentially admitted the fatal flaw: when the experimental material is almost entirely host tissue, there is no isolated “virus” to test as the IV. Instead, it remains embedded within a complex mixture, making it impossible to attribute observed effects to a single, defined cause.

This problem was further addressed by Joel Warren in his accompanying paper, Progress in the Purification of Viruses of Animals. Reflecting on earlier work, including Wendell Meredith Stanley’s crystallization of tobacco mosaic “virus,” Warren noted that the once-prevailing assumption that crystallinity implied purity had been abandoned. He emphasized that significant uncertainty remained regarding whether any “virus” preparation had achieved true purity:

“The point of view became well entrenched that when the “pure” animal virus was obtained it too would have to be crystalline and a single, definable chemical substance. Lively debate (much of which, unfortunately, is unpublished) raged over such matters as the incompatibility between “living” and “crystalline” states or whether a pure nucleoprotein necessarily had to be a crystal. But these oratorical storms of the late 1930’s have subsided and we are no longer lulled by the comforting thought that crystallinity is synonymous with purity. It has been replaced by the more pragmatic attitude of let us wait and see what each animal virus is like when there no longer exists reasonable doubt as to its purity.

Warren went on to acknowledge that, despite ongoing efforts, the goal of obtaining an unequivocally pure “virus” had not been achieved and appeared increasingly difficult:

Although the isolation of an unequivocally pure animal virus has not been accomplished and this objective appears more remote than it did fifteen years ago, considerable progress has been made towards this goal.

No longer do most workers regard animal viruses as closely akin to plant viruses and equally amenable to procedures employed in plant virus isolation.”

He cautioned that even the physical structures observed in experimental preparations could not be confidently identified as the “virus” itself due to the lack of purification:

“As regards structure, it is perhaps fortunate that the application of the ultracentrifuge in virology preceded that of the electron microscope. Otherwise, it is not unlikely that the literature would be filled with a mass of photographs of those bizarre objects found in crude suspensions whose claim to the designation of “virus” reside only in the infectivity of the material. Thus, not only is their purity in doubt, but their antecedents are dubious and it would have required considerable labor to establish their legitimacy.”

Warren considered it “fortunate” that ultracentrifugation arrived before electron microscopy, believing it prevented researchers from prematurely identifying structures in crude mixtures. But spinning samples at high speeds did not solve the purity problem. As Hubert Loring pointed out, high-speed centrifugation cannot separate a presumed “virus” from normal host material—tissue components spin down right alongside the target particles. In other words, ultracentrifugation did not isolate a single entity; it just concentrated a dirty mixture.

Warren himself further elaborated on the problem of contamination, noting that even materials derived from normal, “uninfected” tissues could closely resemble the particles attributed to “viruses,” both in size and physicochemical properties:

“We must digress for a moment here to the problem of contamination of purified virus preparations by materials derived from the host tissue, and in particular, by that group of heterogeneous substances now loosely designated as “normal tissue components.” Granules have been reported from various normal mammalian tissues, allantoic fluid and milk, whose reported sizes of 20 to 200 mμ, bracket those of the animal viruses. The chemical and physical properties of the large bodies are also closely akin to viruses and they contain ribonucleic acid. If these merely represent fragments of larger broken structures (7), then their homogeneity of size and form and heterogeneity of composition are truly remarkable. It seems more likely that some normal components, such as those of brain and allantoic fluid, are entities in themselves whose cellular role is obscure. The nature of these normal components and their relation, if any, to animal viruses is a problem deserving of more investigation.”

This creates an even bigger problem: researchers could not tell normal host cellular debris apart from the presumed “virus” itself. If you cannot distinguish the target particle from host material, you have no idea what you are actually testing. Without a purified particle, there is no independent variable, and without an independent variable, there is no science.

Warren next outlined the principal difficulties hindering the study of “viruses,” emphasizing that experimental preparations remained complex and that both purification and observation were limited in practice. He identified several key issues:

• Purified preparations are contaminated with normal cellular constituents indistinguishable from “viruses.”
• There are not enough “viral” particles in purified samples to “infect” and study.
• Purification leaves the presumed “viruses” ineffective.
• Electron microscopy were operating near their limits of resolution in the relevant size range.

He summarized these limitations as follows:

“What have we learned in the last decade about the particles of that group of small viruses which includes the agents of poliomyelitis, the arthropod-borne encephalitides and yellow fever? It is here that the purification of mammalian viruses has made the least headway, and the reasons for this are simply stated:

1. As already mentioned, preparations have been contaminated by normal components with properties, such as density and sedimentation, often indistinguishable from the infectious agent.
2. In the case of certain diseases, for example poliomyelitis, the infectivity (and probably the number of particles) is so low that considerable concentration must be effected before sufficient virus is available for study.
3. As concentration and purification proceed, the particles of virus tend to aggregate into masses whose sedimentation and infective levels are no longer those of the individual unit.
4. Finally, in the range of 10 mμ (100 angstroms), our present electron microscopes are operating only a little above their limits of resolution.

With all respect for the considerable amount of effort expended on the purification of the viruses less than 40 mμ in size, it seems justifiable to state that thus far there has not been much original, or significant, knowledge contributed to the general nature of the animal viruses by these investigations.”

Many of the “viruses” under investigation at the time, including those associated with poliomyelitis, fell within this sub-40 mμ range, exactly where both purification methods and visualization techniques were operating at their limits. At this scale, normal cellular components like ribosomes, vesicles, and protein aggregates overlap in size, density, shape, and sedimentation behavior. Because these cellular components were practically indistinguishable from one another, there was no way to tell a supposed “virus” apart from normal cell debris, and isolating a clean independent variable was impossible.

Warren noted that further attempts to improve purification techniques continued to encounter the same fundamental limitations. In experiments using protamine treatment followed by ultracentrifugation, researchers initially believed they had obtained more refined “virus” preparations, as indicated by reduced nitrogen content. However, this interpretation was undermined when identical results were obtained from normal, “uninfected” tissue processed in the same manner:

“Any hope that the concentrate obtained from the protamine-treated material was pure virus was dispelled when the same nitrogen content was found in purified sediments from suspensions of normal brain tissue prepared in an identical manner.”

Efforts to visualize these preparations using electron microscopy proved equally inconclusive:

“Attempts at electron visualization of protamine-clarified ultrasediments encountered the same difficulties we have already discussed. So much extraneous debris was present that it was difficult to determine which of the particles was virus.”

Although additional methods, such as enzymatic digestion, were introduced to further reduce extraneous material, even these approaches did not fully resolve the problem of identification. At best, they allowed investigators to “hazard a good guess:”

“This procedure, also applicable to other agents, makes satisfactory electron micrography of the smaller viruses possible and enables one to hazard a good guess as to which particle represents the virus. Graphic evidence of this is seen in figures 1-3 (Plate 1).

Figure 1 is an electron micrograph of encephalomyocarditis virus purified and concentrated in the manner described. The larger particles, measuring approximately 30 mμ, are those of the virus. Also seen in this preparation are occasional smaller (18 mμ) particles of a normal mouse brain component. Particles obtained from normal mouse brain are shown in figure 2. Note the extreme homogeneity of the material. The third electron micrograph contains the Texas type of Coxsackie virus obtained from infected mouse brain. The particles measure approximately 24 mμ and it will be appreciated that the differentiation of the virus from the normal component on the basis of the electron photograph is essentially impossible.”

As Warren noted, even in these more refined preparations, particles derived from normal tissue remained present and overlapped in size with those attributed to the “virus,” making differentiation effectively impossible on visual grounds. This highlighted the impenetrable wall early virology kept hitting: more complex purification protocols still failed to produce a pure agent. The presumed “virus” remained indistinguishable from normal biological material, leaving the independent variable neither definitively isolated nor unambiguously identified.

Warren’s concluding remarks captured how far virology’s goals had moved following decades of failed purification efforts. He noted that the earlier debates concerning the molecular nature or crystallinity of “viruses” had largely receded, and that the isolation of static, purified preparations was no longer sufficient to address the central biological questions:

“A knowledge of the static physical and chemical properties of purified viruses is obviously insufficient in itself to enable us to understand the essential problems of their metabolism and capacity for self-duplication.”

He further cautioned against an exclusive focus on what remained after purification, emphasizing that such preparations may represent only a stripped-down residue rather than the functional entity under investigation:

“It would be a pity if attention were exclusively confined to the bare, flayed residue which can be the result of a successful purification of one of the more stable viruses.”

Instead, Warren reframed the core investigative challenge, steering it away from basic isolation toward biological function and mechanism:

“Can we isolate a dynamic virus system, and from whence comes its energy?”

​Even with refined purification methods, Warren acknowledged that “viral” preparations failed to solve the primary explanatory problems. Instead of focusing on isolating a stable physical entity—a true experimental variable—the field turned toward the broader question of how these inferred agents functioned inside a host.

Purification of the presumed “viral” particles was clearly an ideal to be obtained. But because virologists could not find enough particles in primary patient fluids, and because the harsh chemical processing used to purify them was reported to destroy “infectivity,” a crucial problem emerged. When the material was actually subjected to purification, the resulting preparations failed to retain the very property they were supposed to demonstrate—the ability to cause “infection” or disease. Rather than overcoming this hurdle to isolate a true independent variable, virologists bypassed direct isolation entirely, opting to cultivate samples in tissue cultures. They built an entire framework on contaminated, biological concoctions, leaving the presence of a “virus” in the original sample as an unproven assumption.

The Tissue Culture Compromise

Around the same time as the critical admissions made during the Symposium on Viral and Rickettsial Diseases highlighting the difficulties in purifying the polio “virus,” Hubert Loring and Carlton Schwerdt reported in Isolation of a Macromolecular Constituent with Properties of the Lansing Strain of Poliomyelitis Virus that they had obtained a concentrated preparation associated with the Lansing strain of poliomyelitis “virus.” To do so, the researchers relied on extraction from the brains and spinal cords of 4–6 week old cotton rats experimentally inoculated via intracerebral injection. However, despite efforts to achieve maximal purification, they acknowledged that host-derived macromolecular material remained present in the preparations. Based on comparative nitrogen measurements from normal tissue controls, they estimated that they had achieved at least 80% purity:

“If the normal protein is present in the infectious extracts to the same extent it is found in normal tissue, it is evident that the maximum amount of normal macromolecular impurity present in the purified virus is of the order of 5 to 20%.”

In other words, even under optimized conditions, the resulting preparation still contained a measurable proportion of “non-viral” macromolecular material under their analytical assumptions.

In Schwerdt’s later 1954 paper Morphology of type II poliomyelitis virus (MEFl) as determined by electron microscopy, co-written with Wendell Stanley, the authors noted that earlier electron microscopy studies were limited by uncertainty in identifying which particles corresponded to the “infectious” agent due to the presence of multiple morphologically similar structures. They wrote:

“Previous attempts to determine by electron microscopy the size and shape of particles of human poliomyelitis virus have been equivocal because of uncertainty as to the identity of the virus particle among various objects of similar appearance found in the electron micrographs.”

Improved purification methods applied to “virus” propagated in central nervous system tissue or monkey kidney cell culture reduced the heterogeneity of observable particulate material. The investigators further reported that:

“Improved methods of purification applicable to poliomyelitis virus propagated in central nervous system tissue in vivo, or in monkey kidney tissue culture, now yield virus concentrates containing at most 2 classes of particles and frequently only one.”

They also noted that “infectivity” could be statistically correlated with particle counts in these preparations, which was used as part of the basis for identifying the presumed “viral” particle.

Later historical accounts highlight just how crude early “poliovirus” extraction really was. Past efforts such as those by Loring and Schwerdt were said to only achieve purification of 1% of the polio “virus,” while later work by Schwerdt and Howard Bachrach was claimed to achieve 10% purity. As extraction techniques advanced, researchers were forced to adopt much stricter benchmarks for what actually qualified as a pure sample.

In 1959, Carlton Schwerdt published the paper Purification and Properties of Poliovirus, in which he stated that the physical and chemical study of a “virus” required preparations of the “greatest possible degree of purity.” He then provided the ideal for what exactly a purified “virus” preparation should be—an identical mixture of all of the same particles:

“Studies of the physical and chemical properties of the human polioviruses by direct means require, as is the case with any virus, preparations of the greatest possible degree of purity. This poses the problem of establishing criteria of purity that can be met experimentally. Perhaps the ideal operational concept of a pure virus preparation is one in which all particles are identical with respect to measurable physical properties and each possesses the property of infectivity. This concept may not necessarily be adequate for all viruses, since virus particles of detectable chemical and physical heterogeneity may nevertheless possess the same infectivity.”

Schwerdt noted that attempts to obtain the polio “virus” directly from samples of central nervous system tissue were handicapped by “low virus concentration” as well as substantial contamination from normal macromolecular tissue particles with physical and chemical characteristics similar to those attributed to the “virus.” Researchers were unable to identify a polio “virus” until they began using tissue cultures that were “relatively free of contaminating, nonviral components.” This is an implicit admission that researchers could not identify the independent variable prior to performing tissue culture experiments, which constitutes a direct violation of the scientific method:

“Until recently the usual source of human poliovirus for purification studies was limited to portions of the central nervous system (CNS) of infected primates (Loring and Schwerdt, 1942; Stulberg et al., 1948) or the entire CNS of rodents infected with Type 2 polioviruses (Gard, 1943; Loring and Schwerdt, 1946; Bachrach and Schwerdt, 1952, 1954). Attempts to purify virus from such sources were severely handicapped by the low physical concentration of virus in the CNS tissue, as well as by the presence of large amounts of so-called normal, macromolecular tissue particles of physical and chemical characteristics similar to those of the virus. With the advent of poliovirus propagation in tissue culture (Enders et al., 1949), however, purification studies progressed more rapidly. Tissue culture not only provided an abundant source of virus for purification, relatively free of contaminating, nonviral components (Schwerdt and Schaffer, 1955, 1956), but also a means of accurate assay (Dulbecco and Vogt, 1954), which has served as an aid in following purification steps and in identifying the physical particle with which infectivity is associated (Schwerdt and Fogh, 1957).”

From all of these accounts, it becomes clear that, until the advent of tissue culture, researchers were unable to reliably identify the presumed “virus” within primary samples due to low concentrations and the presence of indistinguishable host-derived material. Only after amplification in artificial systems—where multiple variables are introduced—did they begin to associate specific particles, which had not been independently identified in the original host sample, with “infectivity.”

This sequence inverted standard experimental logic. Instead of isolating an independent variable first and testing it, researchers inferred its presence after observing effects in complex, unpurified mixtures. The cause was never established before the experiment; it was derived from the outcome it was supposed to explain.

A Textbook Recognition of the Sequence

The distinction between the original biological material, tissue culture, and the subsequent purification of the presumed “virus” was not confined to the early virology literature. A much later discussion by Frederick Attenborough, drawing on Jawetz, Melnick & Adelberg’s Medical Microbiology, provides an instructive example of how this sequence was understood in standard medical microbiology.

In discussing the processing of clinical specimens, Attenborough states that:

“those starting materials have to go through a process of ‘purification’ prior to their actual inoculation onto a cell-culture.”

Although Attenborough appears to conflate or rearrange passages from Jawetz concerning the precise order of procedures, his interpretation is nevertheless revealing. He places purification prior to inoculation into cell culture, indicating that he understood the scientifically rigorous sequence to involve separating the presumed agent from its starting material before using it to investigate its biological effects.

This interpretation is consistent with Jawetz’s own discussion of purified “virus.” The textbook states:

“Pure virus must be available in order for certain types of studies on the properties and molecular biology of the agent to be carried out.”

This statement confirms that the starting material obtained from a clinical specimen is not automatically synonymous with “virus.” It is a complex biological mixture containing host-derived material and potentially numerous other components. If that material is introduced directly into a cell culture, any subsequent cellular effect cannot automatically be attributed to a particular particle within the mixture merely because the material originated from a diseased individual.

Jawetz also acknowledges the difficulty of achieving the very condition identified as necessary:

“It is very difficult to achieve complete purity of viruses. Small amounts of cellular material tend to adsorb to particles and copurify.”

Thus, the problem is not simply whether investigators perform a procedure called “purification.” The more important question is whether that procedure actually separates the presumed agent from the surrounding biological material sufficiently to establish what is being studied.

Jawetz then provides an especially important operational description of what constitutes evidence of purity:

“The minimal criteria for purity are a homogeneous appearance in electron micrographs and the failure of additional purification procedures to remove ‘contaminants’ without reducing infectivity.”

A preparation is not demonstrated to be pure simply because it has undergone centrifugation, filtration, chromatography, or another separation procedure. Purity requires evidence that the resulting preparation is sufficiently homogeneous and that further attempts at purification do not reveal additional material without simultaneously removing the property being attributed to the presumed “virus.”

This ties directly back to Schwerdt’s operational definition of a pure “virus:” a sample where every particle shares identical physical properties and every particle is “infective.” In both formulations, the issue is ultimately one of homogeneity: can the material being studied be shown to represent a uniform population of particles possessing the defining property attributed to the agent?

This is the critical methodological sequence. If purification is intended to establish a defined and sufficiently homogeneous experimental material, then the more rigorous sequence suggested by Attenborough’s discussion would be to first separate and characterize the presumed particles from the surrounding biological material and only afterward introduce that preparation into a cell-culture system to determine its effects.

If, instead, the material is first introduced into cells and amplified through the culture system before the presumed agent is purified or characterized, then the culture process itself becomes part of the process by which the agent is identified and defined. This obviously leads to the unavoidable question:

When, exactly, was the independent variable established?

If the proposed cause is supposed to be a discrete biological particle, then that particle should be independently established as such before the experimental system is used to determine what it causes. Otherwise, the experiment uses the effect-producing system itself to help define the very entity presumed to be responsible for the effect.

4. Separation Anxiety

“How can we be sure that we are isolating and quantifying extracellular vesicles rather than enveloped viruses present in the sample? Equally, how can viral researchers know that they are not detecting similarly sized non-viral vesicles or empty vectors during vaccine production?”

https://web.archive.org/web/20201228093419/www.nanoviewbio.com/exosome-blog/2020/5/5/extracellular-vesicles-and-viruses-two-sides-of-the-same-coin

A Century of Absence

By now, it should be clear that the purity standard is not an arbitrary criterion. Its roots trace back to the very birth of bacteriology, and the logic of requiring a verified, isolated causative agent free of confounding factors is inherently tied to the scientific method. Robert Koch and his contemporaries understood that, without demonstrating from the outset that a single species of microorganism existed prior to testing, attributing causation to it would be both logically and scientifically invalid.

This standard was absent from the very moment the concept of a “virus” was introduced and has never been fulfilled since. At no point was a single, verified species of a “virus” identified and isolated prior to experimentation. As a result, the field has operated without a clearly established and purified independent variable throughout its history. Influential virologists such as Thomas Rivers, Albert Sabin, Wendell Meredith Stanley, and Carlton Schwerdt all acknowledged that, without reliable purification, there was no definitive way to study the physical and chemical properties of the proposed agents, nor to conclusively link any specific particle to the disease it was claimed to cause.

Therefore, the issue is rather simple: without first independently identifying and isolating the proposed agent, its causal role cannot be properly tested. In other words, without having Bigfoot on hand, one cannot demonstrate that Bigfoot caused something simply by pointing to indirect evidence that supposedly implicates the mythical creature.

Despite the simplicity of this principle, defenders of virology often argue that the historical inability to purify and isolate “viruses” has been overcome by modern technological advances. However, contemporary research on extracellular vesicles demonstrates that this objection does not withstand scrutiny.

The Modern Confounder: Extracellular Vesicles

Extracellular vesicles (EVs) are nano-sized, membranous structures released by cells into the extracellular space under both normal and “pathological” conditions. These particles are broadly categorized as apoptotic bodies (500–2000 nm), microvesicles (100–1000 nm), and “exosomes” (50–150 nm). EVs present a major challenge for virology because they overlap extensively with so-called “viral” particles in size, density, morphology, and biochemical composition. As a result, they co-sediment during standard purification procedures and cannot be reliably distinguished from “viruses” using conventional techniques.

The 2016 review Extracellular vesicles and viruses: Are they close relatives? summarized the problem:

“Because EVs are produced by virtually all cells, probably every viral preparation is in fact a mixture of virions and EVs. To study their respective functions, it is necessary to separate EVs and virions.”

“Unless more specifically defined, it is currently virtually impossible to specifically separate and identify EVs that carry viral proteins, host proteins, and viral genomic elements from enveloped viral particles that carry the same molecules.”

“A growing body of evidence indicates that cells infected with enveloped or nonenveloped viruses release EVs that contain viral components. Here, we aimed to create awareness that virus preparations may never be pure but rather are contaminated with diverse subpopulations of EVs, and some of these EVs may be either indistinguishable from or very similar to so-called defective viruses.”

In other words, even under controlled laboratory conditions, so-called “viral” preparations are acknowledged to be mixtures rather than isolated entities.

Similarly, an August 2020 study found that separating EVs—particularly “exosomes” and microvesicles—from “virus” particles remains a “considerable hurdle.” Because of their similar size, density, and sedimentation velocity, “virions” and EVs co-sediment in standard isolation techniques. Neither differential centrifugation nor commercial precipitation reagents can reliably separate them, leading the authors to conclude that it is “near impossible to separate EV from virions by biochemical methods.” They emphasized that more carefully validated purification methods are required before the biological functions attributed to these particles can be confidently interpreted.

Mission Impossible

Another study from 2020 described the challenge even more directly, calling it an “almost impossible mission” to separate EVs and “viruses” using conventional methods. The authors noted that these particles are frequently co-pelleted during ultracentrifugation due to their similar dimensions, and acknowledged that, to date, no reliable method exists that can guarantee complete separation:

“Nowadays, it is an almost impossible mission to separate EVs and viruses by means of canonical vesicle isolation methods, such as differential ultracentrifugation, because they are frequently co-pelleted due to their similar dimension [56,57]. To overcome this problem, different studies have proposed the separation of EVs from virus particles by exploiting their different migration velocity in a density gradient or using the presence of specific markers that distinguish viruses from EVs [56,58,59]. However, to date, a reliable method that can actually guarantee a complete separation does not exist.”

Industry sources have echoed these concerns. A September 2021 discussion by Cell Guidance Systems noted that, because of their similar physical properties, “exosomes” and “viruses” utilize the same techniques for purification, namely ultracentrifugation, precipitation, and chromatography. However, it concluded that none of these methods allow for the separation of “exosomes” and “viruses” in order to distinguish one from another.

More recent work continues to reinforce this point. A 2024 study reported that EVs and “virus” particles produced from “infected” cells share similar size and specific gravity, making their separation and purification “controversial.” The authors acknowledged that EVs coexist with “virus” particles and that standard protocols are insufficient for their separation. They emphasized that meaningful biological analysis cannot proceed until these particles are reliably distinguished.

Attempts to resolve this issue using electron microscopy have also proven inadequate. The same study reported that samples contained numerous impurities following ultrafiltration, making visualization difficult. Even when particles could be observed, distinguishing EVs from “virus” particles based on morphology was found to be unreliable, with the authors concluding that such differentiation “might not be possible.”

Bio-Molecular Coronas and Bound Debris

A 2025 review reinforces that these challenges are not relics of early virology, but remain unresolved even with modern techniques. The authors note that extracellular vesicles released from “virus-infected” cells frequently carry “viral” proteins, RNA, and even entire “viral” genomes. In many cases, these EVs are described as functionally similar to “noninfectious viral particles,” and are often produced in quantities exceeding those of so-called “infectious virions.” This completely undermines the claim that any detected particle or genetic material can be uniquely attributed to a distinct “virus,” as similar structures are naturally produced by the cells themselves.

They further acknowledge that commonly used isolation methods based on size and density cannot reliably distinguish between EVs, “viral” particles, and other biological components. Lipoproteins, for example, can co-isolate due to similar physical properties and may even form complexes with EVs, meaning that observed characteristics may not originate from the vesicles themselves.

Even when attempts at purification are made, the resulting particles may retain tightly associated external material, including proteins attributed to the presumed “virus,” such as the infamous “coronavirus” spike proteins:

“Some corona proteins bind tightly to EVs and resist removal, even with rigorous purification…”

This means that what is being studied may not even represent a discrete entity, but rather a particle surrounded by additional material that cannot be reliably removed or accounted for. This has damning implications for interpreting particles identified as “viruses.” If extracellular vesicles can co-isolate with lipoproteins and other biological material that remains tightly bound despite purification, thereby forming a persistent corona, then particles displaying a characteristic corona-like appearance may instead represent EVs associated with a protein or lipoprotein corona rather than a distinct “viral” entity.

The authors emphasize that purity remains a central and unresolved problem:

Key Unmet Challenges in EV Purification

1. Isolating pure EVs is crucial for EV research. Most EV isolation methods that separate EV fractions by size or density may contain viral particles. Although progress has been made in developing purer EV preparations [59], this remains a challenge, particularly for retroviruses, which share similar size, density, and other biophysical properties with many EV subtypes [15]. Reliable techniques to isolate “pure” EVs need to be developed.
2. EVs generated by retrovirus-infected cells commonly carry viral proteins or even fragments of the viral genomes. These EVs can be classified as noninfectious viruses or viral particles. They interact with cells, triggering responses similar to those induced by enveloped viruses. This similarity adds complexity to functional studies, which must account for these EV-mediated effects.
3. Both EVs and viral particles harbor a corona [60]. Interactions between EV and viral coronas may influence their crosstalk. Consequently, loss of the corona during EV isolation and purification may render studies on EV-virion interactions less reflective of their functional interaction in vivo, as the corona may mediate or modulate these interactions.

These findings collectively demonstrate that, even with modern technological advances, it remains impossible to reliably distinguish so-called “viral” particles from extracellular vesicles based on size, density, or morphology. EVs are simply the modern equivalent of the “normal tissue components” described in early virological literature—host-derived particles that cannot be separated from the suspected agent.

This means that the foundational methodological limitation identified decades ago still persists today. The independent variable is not clearly established prior to experimentation. Instead, researchers are working with mixed populations of indistinguishable particles, making it impossible to attribute observed effects to a single, isolated cause.

5. A Matter of Degree

The Operational Ideal vs. Practical Reality

As R. A. Killington noted in Chapter 4: Virus Purification in the Virology Methods Manual, purification requires the physical separation of “virus” particles from “the host cell milieu in which it has grown.” This step is not optional, but foundational: it is necessary for studies in which the “properties or structure of the virus must be distinguished from those of the host cells or culture medium.” In other words, downstream structural and functional claims depend on the ability to attribute observed properties to a defined entity rather than to the surrounding biological material.

Killington pointed out that purity cannot simply be taken on faith—it has to be proven across several benchmarks, from protein-to-particle ratios and gel diffusion to electron microscopy and electrophoresis. The underlying principle is straightforward: researchers cannot make reliable claims about a biological particle’s structure or function until it has been separated from host cell debris.

In practice, however, biological preparations, particularly those derived from “infected” cell cultures or tissue fluids, are not composed of a single, well-defined entity, but of complex and overlapping populations. These include extracellular vesicles, lipoproteins, protein aggregates, membrane fragments, and incomplete or degraded particle forms, many of which share similar size, density, morphology, and physicochemical properties.

As discussed, this problem is not new. Early virologists repeatedly acknowledged that “virus” preparations were impure and that meaningful study of their physical and chemical properties required improved purification. As Carlton Schwerdt later stated, investigations “require, as is the case with any virus, preparations of the greatest possible degree of purity.” He further defined the ideal operational standard as one in which “all particles are identical with respect to measurable physical properties and each possesses the property of infectivity.”

Homogeneity and functional consistency are not optional benchmarks—they are the only way to prove a study is examining a single, defined entity. But as both historical literature and current extracellular vesicle research reveal, achieving those standards in practice remains an unsolved problem. Rather than yielding uniform populations, standard isolation methods routinely produce mixtures of particles with overlapping characteristics that cannot be fully resolved. This is a limitation that has been repeatedly acknowledged throughout the methodological literature.

For example, in Purification of Influenza Virions (2025), it is noted that “a single round of ultracentrifugation can be very effective” for certain applications in which “purity is not an issue,” while simultaneously conceding that this approach “will result in the co-purification of considerable amounts of contaminating material.” Even when more stringent, multi-step density gradient methods are employed “where a greater degree of purity is required,” the outcome remains qualified: “even this approach cannot completely remove contaminants,” and the resulting preparations contain “low but detectable levels of material whose protein composition is suggestive of ribosomes and microvesicles.”

The Cost of Precision

Increasing purity is not without cost. As the same sources emphasize, “although the degree of purity may be very important for some applications…it comes at the cost of experimental time and an increased risk of reducing viral titre due to damage on handling.” ​Purification, then, is never really a matter of achieving complete homogeneity; it is simply a compromise between enrichment, convenience, and material loss. What is produced at each stage is a preparation in which the proportion of target particles is increased while residual, co-purifying components remain. What is not produced is a singular, fully resolved entity composed solely of the particles under investigation.

The core problem is not whether purification is attempted, but how completely it is achieved. If the purpose of purification is to ensure that observed properties can be attributed to a specific, defined entity, then the acceptable level of impurity cannot be treated as a vague or flexible standard. If mixtures of indistinguishable particles remain in a preparation, any observed effect—whether structural, biochemical, or functional—cannot be uniquely attributed to a single entity, and therefore cannot function as direct evidence of its properties. Under these conditions, causal attribution rests on assumption rather than empirical proof.

The standard of purity therefore means only one thing: you have to completely rule out alternative sources for whatever effect you are measuring.

The ≥99% Threshold

This raises a necessary question: how pure must a preparation be in order to support causal or structural claims about a single class of particles?

At minimum, the preparation must approach practical homogeneity, on the order of ≥99%, with any remaining components identified and shown to be irrelevant to the observed effect. This is not an arbitrary threshold, but a practical expression of the requirement that alternative sources of the observed properties be effectively excluded. If even a small fraction of the preparation consists of indistinguishable or functionally similar particles, then the possibility remains that the observed properties arise from those components rather than the entity under investigation.

Think of this like forensic evidence at a crime scene. If the police find a fingerprint and say, “We are 90% sure this belongs to the suspect, but the other 10% might be from a bystander,” that evidence would be tossed out of court. In science, the ≥99% standard serves the same purpose: you must be certain the “fingerprint” (the biological effect) belongs to the “suspect” (the “virus”) and nothing else. Anything less is merely a guess.

This standard mirrors the requirements for the “Pharmaceutical Grade” classification, which mandates ≥99% purity specifically to ensure the absence of fillers, additives, or unknown substances. Regulatory agencies like the FDA or EMA enforce these strict guidelines because they recognize that in a complex mixture, even a 1% unknown fraction can fundamentally alter the quality, performance, and safety of the substance. If we demand this level of purity for the chemicals used to treat a condition, it is only logical to demand the same standard of purity for the biological material used to define the cause of that condition.

What this means is that unless the overwhelming majority of the material can be confidently attributed to a single, defined class of particles, and any residual components are accounted for, the preparation cannot function as a valid independent variable. Instead, it remains a mixture, reducing every subsequent claim from a direct demonstration to indirect guesswork.

The Orthogonality Requirement

This threshold also aligns with the rigorous standards set by the American Chemical Society (ACS) for establishing scientific truth in complex biological and chemical contexts. As noted in the Journal of Medicinal Chemistry, “purity assessment is the logical prerequisite for the accurate characterization of any research grade material,” and it is perhaps “most critical in the case of novel compounds to which a biological activity is ascribed, because trace impurities of high potency can lead to false conclusions.”

​In other words, if a substance, whether a synthetic drug or a presumed biological agent, is not purified to a level where the chemical constitution is accurately known, any reproducibility or interpretability of the research data is fundamentally compromised. By ACS standards, the integrity of a scientific conclusion hinges on whether the material has been characterized with enough precision to exclude the influence of the “host milieu” or trace contaminants that may themselves possess high biological potency.

Central to this ACS standard is the principle of orthogonality, which is the requirement that purity be verified using analytical methods that are mechanistically independent of the process used for isolation. In other words, if a presumed biological particle is “isolated” or “defined” based on its biological activity (such as its ability to induce cytopathic effect in a cell culture), its existence and purity must be confirmed by a completely different, non-biological physical measurement (such as a direct chemical analysis of its unique molecular mass or a specific spectral signature).

Orthogonality is essentially checking your work in a completely different way. If you weigh a bag of apples on a scale and it says 5 lbs, you do not “verify” it by weighing it on the same scale again. You verify it by counting the apples or measuring their volume. In virology, researchers often use the same “scale” (the cell culture) for both the “isolation” and the “proof.” That is not a check; it is simply repeating the assumption.

The “Circular Confirmation” Loop

Without orthogonal validation, researchers risk falling into a loop of circular confirmation, where the biological effect used to define the entity is the only method used to “see” it. If the only way to verify the presence of a “virus” is through the very system of culture and “infection” used to generate it, the “purity” of that preparation remains a theoretical claim rather than a demonstrated physical fact. True isolation requires that the physical entity be established as an independent variable through means that are entirely independent of the experimental effects it is supposed to cause.

​Thus, an independent variable cannot be a artifact of the experiment; it must exist beforehand. This creates a clear methodological dilemma: is the entity identified in the original host, or merely inferred after laboratory propagation?

​Defenders of standard culture methods often claim that isolating “viral” particles directly from host fluids is impractical because “viruses” require a host cell to replicate. But this conflates replication with physical presence. By virology’s own definition, the “virion” is the fully formed, extracellular particle existing outside the cell—supposedly present in massive quantities within patient fluids. If these particles exist in high numbers, they should be physically identifiable directly within host tissue without relying on culture amplification first. If they cannot be separated from host structures prior to lab manipulation, their existence remains tied to the culture system itself.

​Lab cultures do not just host the sample; they alter it. Tissue systems introduce stress, cellular debris, extracellular vesicles, and broken particles, alongside added variables like fetal bovine serum, nutrients, and antibiotics. Any particles observed after culturing reflect a mix of original material and process-induced artifacts.

​Starting with cultured material mistakes the laboratory process for the agent itself. At a minimum, the proposed agent must be isolated directly from the host first. Without that baseline, its identity rests on what happens inside a lab dish rather than proof of a distinct, standalone entity.

6. Still Missing

Since its inception, virology has faced an independent variable problem. “Viruses” were initially presumed to exist within samples based on the ability to produce artificial, laboratory-created disease in plants and animals. Rather than critically examining the experimental methods, which relied on impure materials, crude filtrates, and mechanical injury to produce these effects, the resulting observations were incorporated into the emerging concept of the “filterable virus.”

Yet the “filterable virus” itself remained absent. It was defined only in negative terms—by what it was not—rather than by the direct identification of a distinct entity. It remained an enigma precisely because researchers could not determine what it was they were supposedly working with.

Presupposing the Entity

Although early investigators repeatedly attempted to obtain the equivalent of a “pure culture” for their studies, as has been demonstrated, this goal was never achieved. As a result, virology increasingly relied on indirect lines of evidence to infer the presence of the unseen entity. These included cytopathic effects in tissue and cell cultures, electron microscopy images obtained from impure preparations, detection of supposed “antibodies,” identification of “viral” proteins, and the reconstruction of genome sequences.

However, each of these approaches is a downstream method whose interpretive validity depends on a prior step: the purification and isolation of the particles they are meant to detect. Without first identifying the presumed “viral” particles as a distinct and purified entity, these methods cannot independently establish their existence. Instead, they merely interpret signals produced within complex mixtures whose components have not been fully resolved.

Without this crucial step of establishing the independent variable prior to experimentation, the resulting evidence cannot function as direct demonstration. It becomes inferential by definition.

Defenders of virology frequently point to these converging lines of evidence as collective proof of the agent’s existence. Yet because the independent variable was never isolated first, each line of evidence remains inferential and assumption-based. Cytopathic effects, electron micrographs, “antibody” assays, “viral” proteins, and reconstructed genomes may yield signals consistent with the hypothesis, but interpreting those signals as “viral” presupposes the very entity they are meant to demonstrate.

From the earliest “filterable-virus” experiments to modern genomic reconstructions, the pattern has remained the same: the entity is assumed, traces are observed, and conclusions are drawn. Over time these indirect observations have accumulated into what appears to be a coherent body of evidence, yet the foundational step required to establish the independent variable has never been completed.

Traces Without a Source

The situation is analogous to attempting to prove the existence of a creature like Bigfoot using scattered traces: random droppings in the forest, stray hair fibers, partial footprints, blurry video footage, and distant, unexplained sounds. Such clues might appear suggestive, but their interpretation depends entirely on first establishing that the creature itself exists. Without the creature in hand, there is no way to verify that any of the collected traces actually belong to it.

In the case of virology, the same methodological problem remains. The supposed traces are repeatedly presented as evidence of the entity, yet the entity itself has never been independently demonstrated as a isolated, identifiable variable.

More than a century of research has done nothing to close this gap.

Bigfoot is still missing—and so too is the independent variable in virology.