Field · Emerged 1892 – 2005

Virology

What is an agent that passes through a filter which stops all bacteria, multiplies only inside living cells, and can be crystallised like a salt?

4 chapters5 min read6 turning points 1 open problem

Branched from: Microbiology
Branched into: Not yet surveyed past here
Figures: Dmitri Ivanovsky, Martinus Beijerinck, Frederick Twort, Félix d'Hérelle, Wendell Stanley, Frederick Bawden, Norman Pirie, John Enders, Thomas Weller, Frederick Robbins, Françoise Barré-Sinoussi, Luc Montagnier, Robert Gallo, Jeffery Taubenberger, Johan Hultin

In brief

Bacteriology had a method: pass the fluid through a porcelain filter fine enough to hold back every known microbe, and whatever remains infectious is a dissolved poison. In 1892 Dmitri Ivanovsky found that the sap of diseased tobacco plants stayed infectious after filtration, and assumed a toxin or a flawed filter. Martinus Beijerinck repeated the work in 1898, showed that the agent multiplies in growing tissue and cannot be grown on any medium at all, and called it a contagium vivum fluidum — a living infectious fluid.

What viruses actually are took another forty years. Wendell Stanley crystallised tobacco mosaic virus in 1935, which is behaviour expected of a protein and not of an organism. Electron microscopes then showed particles of fixed size and shape, and chemistry showed each to be a nucleic acid genome inside a protein shell. A virus is not a very small cell; it is a set of instructions with a delivery mechanism, carrying no metabolism of its own and reproducing only by commandeering machinery that belongs to something else.

Key ideas

Filterable agentEnters 1892 – 1898: An infectious agent small enough to pass a filter that retains bacteria — pores around 0.2 µm against bacteria around 1 µm. Filterability was the operational definition of a virus for forty years, before anyone could see one.
Obligate intracellular replicationEnters 1892 – 1898: A virus has no metabolism. It cannot make protein, generate energy or copy itself without a host cell's ribosomes, enzymes and nucleotides, which is why no medium of any composition will grow one.
BacteriophageEnters 1915 – 1917: A virus of bacteria. Phages lyse their hosts, which makes them visible as clear plaques in a bacterial lawn and countable one by one — the feature that made them the workhorse of early molecular biology.
VirionEnters 1935 – 1939: The particle: a nucleic acid genome packaged in a shell of protein subunits, often with a lipid envelope stolen from the host membrane. Being a definite macromolecular assembly rather than a cell, it can be purified, weighed and crystallised.
Plaque assayEnters 1949: Growing virus in cultured cells produces countable holes of dead cells, so a sample's infectivity can be quantified. It turned virology from a descriptive science into one that could titrate, dilute and compare.
QuasispeciesEnters 1983 – 1987: An RNA virus population is not a clone but a cloud of related genomes, because its polymerase makes errors at roughly one per genome per replication. The cloud, not any one sequence, is what selection acts on, and it is why resistance appears so readily.

Draws on other domains

↙ Mathematics
Group Theory
Why so many viruses are icosahedra

Chapter I

What the Filter Let Through

By 1890 bacteriology had a reliable way to tell an organism from a poison. Pass the infectious fluid through unglazed porcelain with pores around 0.2 micrometres; bacteria, which are around a micrometre across, are retained. If the filtrate still causes disease, the cause is a dissolved chemical.

Dmitri Ivanovsky applied this to tobacco mosaic disease in 1892 and got the wrong answer by the book. The filtrate was infectious, so he inferred a bacterial toxin, or a filter that had failed. Six years later Martinus Beijerinck, apparently unaware of the Russian paper, did the experiment that distinguishes the possibilities. He took filtered sap, infected a plant, took sap from that plant, filtered, infected again, and repeated. A fixed quantity of toxin would be diluted away. The infectivity did not weaken. Whatever it was, it was multiplying — and it would not grow on any medium he could devise, only in living tissue. He called it a contagium vivum fluidum.

In the same year Loeffler and Frosch showed that foot-and-mouth disease has a filterable cause, so the phenomenon was not a quirk of plants. Frederick Twort and Félix d'Hérelle then found, independently, that bacteria have their own viruses. These bacteriophages clear a culture overnight and leave countable holes — plaques — in a lawn of bacteria on agar, each one the progeny of a single particle. A virus you can count, grow in a day and mutate at will became the organism of choice for early molecular biology, which is why so much of what is known about genes was learned from phage.

Chapter II

Crystals, Cultures and the Question of Life

Wendell Stanley processed tonnes of infected tobacco leaves and in 1935 obtained needle-like crystals which, redissolved, still caused disease. Crystallisation is what a pure chemical substance does. Stanley called the material a protein; Frederick Bawden and Norman Pirie found two years later that it also contained RNA, which is in fact its genome. By 1939 electron micrographs showed rods of a single size.

This is where the question "is a virus alive?" stops being interesting and starts being badly posed. A virion has no metabolism, makes no ATP, synthesises no protein, and can sit dry on a surface for months. Inside a cell it is an actively evolving population with ancestry, variation and selection. The honest answer is that the categories were built for cells.

John Enders, Thomas Weller and Frederick Robbins made virology a quantitative laboratory science in 1949 by growing poliovirus in cultures of ordinary human embryonic tissue, overturning the belief that it needed nerve cells and animals. Virus could then be produced in bulk, measured by plaque count, and weakened by repeated passage through unnatural hosts — the three requirements for a vaccine. Salk's inactivated vaccine came in 1955 and Sabin's attenuated one soon after.

Chapter III

A Closer Look: How a Dilution Proves Something Is Alive

Return to 1898, because Beijerinck's serial passage is the whole argument, and it is worth doing with numbers: the conclusion is not obvious, and it was reached without a microscope.

Suppose the diseased sap contains a toxin at a generous 1 milligram per millilitre. At each passage, a drop is transferred into a fresh plant and the agent is recovered in new sap — call it a thousandfold dilution of anything that is not replicating. After five passages the dilution factor is

(10^{3})^{5} = 10^{15},

so the toxin concentration would be $10^{-15}$ mg/mL, or $10^{-18}$ g/mL. For a protein toxin of molar mass 50,000 g/mol, that is

\frac{10^{-18}}{5 \times 10^{4}} = 2 \times 10^{-23} \text{ mol/mL},

and multiplying by Avogadro's number gives about 12 molecules per millilitre. Twelve molecules cannot disease a plant. If the fifth-passage sap is as infectious as the first, the agent is being made anew in each host. That is the experiment's logic, and it required no microscope, no chemistry, and no idea of what a virus is.

The sizes involved explain why nobody saw one for another forty years. A Chamberland filter stops particles above roughly 200 nm. Tobacco mosaic virus is a rod 18 nm across and 300 nm long — it passes because it is thin, not because it is short. Poliovirus is a 30-nm sphere, influenza about 100 nm, and the largest, the mimiviruses found in 2003, reach 750 nm and are retained by the same filters that defined viruses in the first place. The operational definition ran out, as operational definitions do.

Visible light cannot resolve any of this. The diffraction limit puts the floor at about 200 nm, so every virus but the giants is below the resolution of any optical microscope ever built. Virology had to wait for the electron microscope, which is why the field spent its first half-century reasoning about an entity none of its practitioners had seen.

Chapter IV

Error Rates and Their Consequences

The property that makes RNA viruses so difficult is arithmetical. Their polymerases lack proofreading, and mutate at around $10^{-4}$ per nucleotide per replication. For a 10,000-base genome that is about one mutation per genome per copy, so a virus population is never a clone — it is a cloud of related sequences, a quasispecies, with the cloud as the unit selection acts on.

For HIV the consequence is immediate: an untreated patient produces on the order of $10^{10}$ virions a day, so every single-base change compatible with viability exists somewhere in that patient by the end of the week. Any drug whose resistance requires one mutation is therefore already defeated before it is given. The arithmetic of how many drugs must be combined to outrun this is worked through under pharmacology, and it is the reason HIV treatment is always a combination.

The same reasoning, pointed at populations instead of patients, is what infectious disease dynamics is about: a virus that mutates fast within hosts also evolves fast between them, which is why influenza vaccines are reformulated annually and why variant surveillance is now a permanent function of public health. And the 1918 genome, reconstructed from fixed autopsy tissue and a body in Alaskan permafrost, turned out to carry no human-adapted segments at all — the pandemic began with a bird virus that learned to transmit, which is also the scenario that the biosafety argument of the last twenty years is about.

Turning points6

1892 – 1898DISCOVERYContested
An infectious agent that passes the filter
Dmitri Ivanovsky, investigating tobacco mosaic disease, finds that sap from diseased plants remains infectious after passing a Chamberland filter that stops all bacteria. He suspects a toxin or a defective filter. Martinus Beijerinck, independently, shows in 1898 that the agent multiplies in growing tissue — it can be serially passaged through many dilutions without losing strength — and that it will not grow on any laboratory medium. In the same year Loeffler and Frosch show that foot-and-mouth disease has a filterable cause, the first animal virus.
What is disputed
Priority is still argued, and the argument is partly about what counts as discovery. Ivanovsky published the observation six years earlier but interpreted it as a bacterial toxin or a technical failure; Beijerinck drew the correct conclusion that the agent was a new kind of entity, and seems not to have known Ivanovsky's paper. Russian and Dutch accounts weigh the two differently.
Dmitri Ivanovsky, Martinus Beijerinck
› Sources (3)
- Ivanovsky, D. (1892). Über die Mosaikkrankheit der Tabakspflanze. Bulletin Scientifique de l'Académie Impériale des Sciences de St.-Pétersbourg 35: 67–70.
- Beijerinck, M. W. (1898). Über ein Contagium vivum fluidum als Ursache der Fleckenkrankheit der Tabaksblätter. Verhandelingen der Koninklijke Akademie van Wetenschappen te Amsterdam 65: 1–22.
- Bos, L. (1999). Beijerinck's work on tobacco mosaic virus: historical context and legacy. Philosophical Transactions of the Royal Society B 354: 675–685.
1915 – 1917DISCOVERY
Viruses that eat bacteria
Frederick Twort notices that some of his staphylococcal cultures become transparent, and that the clearing agent passes a filter and can be transferred indefinitely to fresh cultures. Félix d'Hérelle, studying dysentery, finds the same phenomenon, names it bacteriophage, and recognises that each clear plaque in a bacterial lawn is the progeny of one particle. Phages gave biology a virus that could be counted, grown overnight and studied genetically, and they became the material of the molecular biology that followed.
Frederick Twort, Félix d'Hérelle
› Sources (2)
- Twort, F. W. (1915). An investigation on the nature of ultra-microscopic viruses. The Lancet 186: 1241–1243.
- d'Hérelle, F. (1917). Sur un microbe invisible antagoniste des bacilles dysentériques. Comptes Rendus de l'Académie des Sciences 165: 373–375.
1935 – 1939DISCOVERY
A virus in crystals
Wendell Stanley purifies tobacco mosaic virus from tonnes of infected leaves and obtains needle-shaped crystals that remain infectious. A crystal is what a chemical substance forms, not what an organism does. Stanley reported the material as a protein; Frederick Bawden and Norman Pirie showed in 1937 that it also contains RNA, which is the genome. In 1939 electron micrographs showed uniform rods about 18 nm across and 300 nm long. A virus is a definite molecular object.
Wendell Stanley, Frederick Bawden, Norman Pirie
› Sources (3)
- Stanley, W. M. (1935). Isolation of a crystalline protein possessing the properties of tobacco-mosaic virus. Science 81: 644–645.
- Bawden, F. C. & Pirie, N. W. (1937). The isolation and some properties of liquid crystalline substances from solanaceous plants infected with three strains of tobacco mosaic virus. Proceedings of the Royal Society B 123: 274–320.
- Creager, A. N. H. (2002). The Life of a Virus: Tobacco Mosaic Virus as an Experimental Model. University of Chicago Press.
1949TECHNIQUE-INVENTED
Growing virus without animals
Poliovirus was thought to grow only in nerve tissue, which meant propagating it in monkeys — expensive, variable and unsuited to making a vaccine. John Enders, Thomas Weller and Frederick Robbins grow it in cultures of human embryonic skin and muscle, with antibiotics to suppress bacteria, and detect it by the damage it does to the cells. Virus could now be produced in quantity, titrated by counting plaques, and attenuated by repeated passage. The Salk and Sabin vaccines followed within six years.
John Enders, Thomas Weller, Frederick Robbins
› Sources (2)
- Enders, J. F., Weller, T. H. & Robbins, F. C. (1949). Cultivation of the Lansing strain of poliomyelitis virus in cultures of various human embryonic tissues. Science 109: 85–87.
- Oshinsky, D. M. (2005). Polio: An American Story. Oxford University Press.
1983 – 1987DISCOVERYContested
The virus behind AIDS
Three years after the first reports of a new immune deficiency, Françoise Barré-Sinoussi and Luc Montagnier isolate a retrovirus from the lymph node of a patient and publish it in 1983. Robert Gallo's laboratory publishes a virus it calls HTLV-III in 1984, with a blood test, and claims the causal link. The viruses turned out to be the same isolate, the French sample having contaminated the American laboratory. A test soon protected the blood supply, and the virus's high mutation rate shaped every attempt at treatment and vaccine that followed.
What is disputed
The priority dispute was resolved by an agreement between two heads of state in 1987, splitting patent royalties, rather than by scientific adjudication. The 2008 Nobel Prize went to Barré-Sinoussi and Montagnier and not to Gallo, a decision still argued over; Gallo's laboratory established the causal connection to AIDS and developed the blood test, while the isolate was French.
Françoise Barré-Sinoussi, Luc Montagnier, Robert Gallo
› Sources (3)
- Barré-Sinoussi, F. et al. (1983). Isolation of a T-lymphotropic retrovirus from a patient at risk for acquired immune deficiency syndrome (AIDS). Science 220: 868–871.
- Popovic, M., Sarngadharan, M. G., Read, E. & Gallo, R. C. (1984). Detection, isolation, and continuous production of cytopathic retroviruses (HTLV-III). Science 224: 497–500.
- Enserink, M. (2008). Nobel Prize: a glaring omission. Science 322: 174.
1997 – 2005TECHNIQUE-INVENTEDContested
Sequencing a pandemic that ended in 1919
Jeffery Taubenberger's group recovers influenza RNA fragments from formalin-fixed autopsy lung tissue held in a US military pathology archive, and from a body buried in Alaskan permafrost recovered by Johan Hultin. Over eight years the complete genome of the 1918 pandemic strain is assembled, showing it to be an avian-descended H1N1 with no reassortment from a human strain — which answered how the deadliest pandemic of the century began, and started a long argument about what should be done with such a sequence.
What is disputed
Reconstructing the live 1918 virus in 2005, and later work making avian H5N1 transmissible between mammals, divided the field over whether the knowledge gained about pandemic potential justifies creating the agent. The disagreement produced a US funding moratorium in 2014, a framework for reviewing such work, and no settled consensus about where the line falls.
Jeffery Taubenberger, Johan Hultin
› Sources (3)
- Taubenberger, J. K., Reid, A. H., Krafft, A. E., Bijwaard, K. E. & Fanning, T. G. (1997). Initial genetic characterization of the 1918 'Spanish' influenza virus. Science 275: 1793–1796.
- Taubenberger, J. K. et al. (2005). Characterization of the 1918 influenza virus polymerase genes. Nature 437: 889–893.
- National Science Advisory Board for Biosecurity (2016). Recommendations for the evaluation and oversight of proposed gain-of-function research. NSABB, Bethesda.

Applications

Where it is used

Gene therapy
Viruses as delivery vehicles
A virus is a machine for getting nucleic acid into a cell, which is precisely what gene therapy needs. Adeno-associated virus and lentivirus vectors, stripped of the genes that let them replicate, now deliver working copies of genes for inherited retinal disease, haemophilia and spinal muscular atrophy, and carry the genes that reprogramme T cells against leukaemia.
› Sources (1)
- Naldini, L. (2015). Gene therapy returns to centre stage. Nature 526: 351–360.
Oceanography
The most abundant biological entities on Earth
Seawater holds roughly $10^{7}$ virus particles per millilitre, some $10^{30}$ in the ocean as a whole, and they kill an estimated 20 to 40% of marine bacteria every day. That lysis releases dissolved organic carbon back into the water rather than passing it up the food chain — the "viral shunt" — which makes viruses a term in the global carbon budget.
› Sources (1)
- Suttle, C. A. (2007). Marine viruses — major players in the global ecosystem. Nature Reviews Microbiology 5: 801–812.
Molecular biology
The tools came from phage
Restriction enzymes are bacterial defences against phage; reverse transcriptase came from retroviruses; CRISPR is an adaptive immune system against phage; T7 polymerase and phage display both come from the same source. The reason so much of the standard laboratory toolkit is viral or anti-viral is that bacteria and their viruses have been in an arms race for billions of years, and the weapons are precise.
› Sources (1)
- Salmond, G. P. C. & Fineran, P. C. (2015). A century of the phage: past, present and future. Nature Reviews Microbiology 13: 777–786.

Open problems

Where the map runs out

Open

Where viruses came from

Open as of 2026; the three classical hypotheses are all still defended, and may each be true of different viruses.

Three accounts compete. Viruses may be escaped fragments of host genomes that acquired coats; they may be descendants of cells that lost everything but their genomes; or they may predate cells, as replicators from the era before the last common ancestor. The giant viruses found since 2003, with genomes larger than some bacteria and hundreds of genes of unknown origin, have been read as evidence for each.

Why it is hard: Viruses leave no fossils and their sequences evolve fast enough to erase deep signal, so phylogenetic reconstruction fails beyond a certain depth. They also share no universal gene — no equivalent of the ribosomal RNA that roots the tree of cellular life — so there may be no single answer to find.
What resolving it unlocks: Whether the virosphere is a by-product of cellular life or a parallel lineage as old as it, which bears on the origin of life itself and on how much of the genome of any organism arrived from outside.

› Sources (2)

Koonin, E. V., Senkevich, T. G. & Dolja, V. V. (2006). The ancient Virus World and evolution of cells. Biology Direct 1: 29.
Krupovic, M., Dolja, V. V. & Koonin, E. V. (2019). Origin of viruses: primordial replicators recruiting capsids from hosts. Nature Reviews Microbiology 17: 449–458.