The Art of Scientific Investigation

1. 1Separate the two phases: load your mind by studying the problem thoroughly, then deliberately step back so relevant information can come to the focal point of thinking without being crowded by excess detail.
2. 2Read critically rather than passively. When you take in the literature, strive to maintain independence of mind and treat what you read as a stimulus to thinking, not a fixed pattern to conform to.
3. 3Widen your inputs deliberately: read outside the immediate problem, outside your own field, outside science. Variety stimulates freshness of outlook; too constant study of a narrow field predisposes to dullness.
4. 4Manage reading load by skim-reading over a wide field, relying on summaries and reviews for material outside your immediate interest so time is economised and only special-interest parts get close study.
5. 5Stay in active research while you read so freshness and originality do not suffer; combine reading as stimulus with hands-on work on the problem.
6. 6Use discussion with others as a stimulus to the mind — put yourself in contact with colleagues so ideas whose connection was not previously suspected can surface.
7. 7Protect a background of fundamental knowledge; superficial skim-reading only pays off if you have foundations solid enough to quickly assess new work and grasp significant findings.

Bessemer, who discovered the method of producing cheap steel, said he had an immense advantage over others working the same problem because he had no fixed ideas derived from long-established practice to control and bias his mind, and did not suffer from the general belief that whatever is, is right. His mind was free of the conditioning that close study of conventional practice imposes — the very handicap Kettering described when he said that from studying conventional textbooks we fall into a rut, and to escape from it takes as much effort as to solve the problem. Yet Bessemer, like most productive 'outsiders', paired freedom from established patterns in one field with knowledge and training in others. The favorable condition here is not ignorance itself but freedom from conditioned thinking held together by real competence elsewhere. This is why Beveridge advises reading critically and keeping independence of mind rather than avoiding reading altogether.

• If knowledge in the field is still growing and the problem is new, then read deeply — all the advantage is with the well-informed expert; if the field is worked out and needs a revolutionary approach, then guard against conditioned thinking and expect the fresh idea from the outsider.
• If information you hold on a problem may be false, then treat it as a serious barrier and re-examine it, because false information is a worse obstacle to new ideas than mere gaps.
• If you are reading outside your immediate interests, then read superficially via summaries and reviews; if it bears directly on your problem, then read closely and critically.
• If your knowledge is narrowing to your speciality, then deliberately broaden inputs (teach, read history of science, attend conferences) to regain perspective — to see the wood rather than the trees.
• If you find yourself seeing the problem exactly as previous writers framed it, then treat that as a signal you have been conditioned and seek a different approach.

• Reading passively and absorbing the field's conventional framing, so the mind is conditioned to see the problem the same way as everyone before — the tell is that you cannot conceive an approach different from published ones.
• Loading the mind with detailed technical information instead of general principles, so you see only the trees — the tell is that you can recite facts but cannot get perspective on the whole problem.
• Confining reading to the problem or one's own field, producing dullness and ever-narrowing knowledge — the tell is that your outlook and interests keep shrinking to your speciality.
• Skim-reading without a background of fundamental knowledge, so you cannot assess new work or grasp significant findings — the tell is that fast reading yields no useful judgment of what matters.
• Treating general principles as fixed laws rather than as flexible understanding — the tell is rigidity that blocks the very originality favorable conditions are meant to produce.

• ▸Problem/field maturity: is knowledge still growing or worked out? (deep-read vs. guard against conditioning)
• ▸Core literature read closely: list, and note where you may be absorbing others' framing
• ▸Suspect information: which held facts might be false or unverified?
• ▸Wide/superficial reading log: summaries and reviews scanned outside your field
• ▸Independence check: one approach you conceived that differs from published ones
• ▸Foundations status: fundamental grounding solid enough to assess new work? (yes/no)
• ▸Perspective check: are you seeing the wood or only the trees right now?
• ▸Discussion partners: who have you talked the problem over with as a stimulus?
• ▸Active-research link: current hands-on work keeping reading as stimulus, not substitute

1. 1Work at the bench for long hours and try novel procedures, not just conventional ones — this maximizes your exposure to the rare significant clue, which is the only aspect governed by sheer chance.
2. 2While watching for the expected result, hold a parallel channel of attention open for the unexpected; treat any deviation from expectation as a candidate clue rather than a nuisance.
3. 3When you notice something anomalous — a culture growing more luxuriantly, colonies inhibited, an unplanned contamination — stop and register it explicitly instead of discarding it.
4. 4Prepare your mind in advance by reading widely and correlating what you read with your own knowledge, so that when a clue appears you have the background to grasp its possible significance.
5. 5Interpret the clue by actively relating it to other knowledge and by asking what further knowledge it could be used to derive — significance is not intrinsic; you confer it.
6. 6Separate observation from interpretation, especially when the observation comes from lay informants; enquire tactfully to establish exactly what was seen before assigning meaning.
7. 7Follow the clue up experimentally rather than merely recording it — Fleming's advantage over those who saw the same thing was perspicacity in seizing the opportunity.

In 1919 J. T. Edwards noticed that one culture in a group of Brucella abortus grew far more luxuriantly than the others and that it was contaminated with a mould. The observation itself carried no intrinsic significance — luxuriant growth plus a mould could easily be dismissed as spoiled or contaminated material, exactly the reaction Scott had to the mould inhibiting staphylococcal colonies, which he considered only a nuisance. The acute observer's move is to notice the anomaly while watching for expected culture behaviour, then refuse to write it off. Fleming's discovery of penicillin turned on precisely this: several people now recall having seen mould inhibiting staphylococcal colonies before him, but only he followed it up and gave the observation significance by relating it to further knowledge. The difference was not luck in seeing the clue but the prepared mind that interpreted it and the perspicacity to seize the opportunity others let pass. This is why Beveridge insists chance favours the alert: the event was available to many, the discovery to one.

• If an event deviates from your expectation, then treat it as a possible clue and investigate before dismissing it as a nuisance or an error.
• If you find something significant across several aspects you started, then concentrate your effort on that aspect rather than staying dispersed on a wide front.
• If a lay informant reports an observation (a gamekeeper, a farmer), then enquire tactfully to separate what they observed from how they interpreted it before acting.
• If the field is still growing and the problem is new, then depth of expert knowledge helps; if the field is worked out, then suspect that established knowledge is handicapping you and seek a fresh outlook.
• If a clue's significance is unclear, then try relating it to other knowledge and ask what further knowledge it could yield — significance is conferred, not found.

• Noticing the clue but failing to appreciate its significance — the tell is irritation: like the physicist annoyed by X-ray evidence before Rontgen, or Scott calling the mould 'only a nuisance'.
• Confining activity to conventional procedures, which minimizes exposure to fortunate accidents — the tell is that nothing unexpected ever seems to happen at your bench.
• Passively taking in data instead of actively hunting for gaps, inconsistencies, and analogies — the tell is that you never formulate hypotheses from what you read or observe.
• Accepting an informant's or author's interpretation as fact — the tell is that you cannot state which part is the reported observation and which is the interpretation.
• Letting minds untrained in discipline notice only events that support a preferred view and forget the rest — the tell is confirmation of prior notions with no anomalies recorded.

• ▸Observed event: state exactly what you saw, stripped of interpretation.
• ▸Expected result: what you were watching for at the time.
• ▸Deviation: how the observation departs from the expected.
• ▸First reaction: note whether you were tempted to dismiss it as nuisance/error.
• ▸Related knowledge: what existing facts or analogies does this connect to?
• ▸Possible significance: what could this mean or what further knowledge could it yield?
• ▸Source reliability: if from an informant, separate their observation from their interpretation.
• ▸Follow-up experiment: the crucial test you will run to pursue it.

1. 1Survey all relevant literature early using standard indexing journals (Quarterly Cumulative Index Medicus, Zoological Record, Index Veterinarius, Bibliography of Agriculture) or a trained librarian; missing even one significant article wastes later effort.
2. 2Collect firsthand local data before touching the lab bench — for an animal disease, run field observations and interview farmers, separating their observations from their interpretations by tactful, searching enquiry.
3. 3Study 'things caused' before causes: define the condition by its manifestations so you can distinguish it from confusable conditions (as Hughlings Jackson advised).
4. 4Marshal and correlate all data, then break the problem into several formulated questions, watching for gaps, contradictions between writers, and inconsistencies between local observation and prior reports.
5. 5Spend long preliminary thought designing crucial experiments — ones giving a result consistent with one hypothesis and inconsistent with another — before starting bench work.
6. 6Work within your technical capacity and resources; open on several aspects at once but concentrate as soon as one proves significant, rather than dispersing effort on too wide a front.
7. 7Before reporting, make a rigorous test for error — measure yourself by how fast you detect and discard your own mistakes.
8. 8Assess results with statistical soundness: learn enough biometrics (Snedecor, Bradford Hill, or Topley and Wilson's chapter) to know when to consult a biometrician, avoiding both undue neglect and undue respect.

The discovery that ferrets are susceptible to canine distemper began as an assertion from a gamekeeper — firsthand lay observation that scientists at first did not take seriously. Rather than dismissing it permanently, they later decided to test whether there was anything in it, and the claim proved important. This illustrates the sequence the passages prescribe: collect firsthand local information, apply tactful enquiry to separate observation from interpretation, then subject the claim to experimental test. Contrast this with Noguchi, who isolated a spirochaete from leptospiral jaundice cases and reported it as the cause of yellow fever — a failure to first define the condition by its manifestations and distinguish it from confusable ones, which delayed yellow fever investigation. The gamekeeper case shows objectivity toward unglamorous sources; the Noguchi case shows the cost of skipping 'the study of things caused' before causes.

• If even one significant article on your problem could exist, then complete a thorough literature survey early rather than after starting bench work.
• If a lay informant reports something that supports their own notions, then use searching enquiry to separate observation from interpretation before trusting it — but do not dismiss it outright (the gamekeeper rule).
• If an experiment would give the same result under competing hypotheses, then redesign it until the result is consistent with one and inconsistent with the other (crucial experiment).
• If one aspect of the work turns up something significant, then concentrate there and stop dispersing effort across the wider front.
• If you cannot judge whether a result is statistically sound, then consult a biometrician rather than neglecting or over-trusting the statistics.

• Undertaking laboratory work with little relation to the real problem because field observation and local enquiry were skipped — the tell is lab findings that don't match how the disease actually occurs locally.
• Studying causes before defining the thing caused, as with Noguchi reporting a leptospiral spirochaete as yellow fever's cause — the tell is confusing your target condition with another it resembles.
• Frantic, ill-conceived experimental activity producing volumes of 'what-of-it' literature (Zinsser's ant-like agitation) — the tell is many experiments run without prior intellectual incubation of the variants.
• Excessive caution to avoid all error, which prevents discoveries as much as mistakes — the tell is a hypothesis birth rate so low nothing new is ever ventured ('panic of error is the death of progress').
• Reporting findings before a rigorous test for error — the tell is slow detection of your own mistakes; skill lies in the speed with which errors are thrown out.

• ▸Literature survey complete: list indexing journals/librarian consulted and confirm no significant article missed.
• ▸Firsthand local data: field observations made and informant statements recorded, separating observation from interpretation.
• ▸Condition defined: manifestations listed and distinguished from confusable conditions.
• ▸Formulated questions: the specific questions the problem has been broken into.
• ▸Competing hypotheses: state the two the experiment must discriminate.
• ▸Crucial-experiment design: how the result will be consistent with one hypothesis and inconsistent with the other.
• ▸Resource/capacity check: confirm the experiment is within your technical means.
• ▸Error test: the rigorous check for error to run before reporting.
• ▸Statistical soundness: biometric method chosen or note that a biometrician will be consulted.

1. 1Sustain active bench work and dabble in novel procedures — the raw frequency of chance opportunities rises with hours at the bench and with unfamiliar techniques that produce unexpected behavior. Manufacture more exposure rather than waiting.
2. 2Adopt an opportunist stance (Dale's term): give attention to every unexpected happening or observation, and investigate those that seem promising. Do not pass over a slight, disconnected point with a half-considered explanation.
3. 3Treat non-reproducible or 'wrong' results as signals, not nuisances. When results vary though known factors are unchanged, suspect an unrecognized factor — but first rule out technical error, since that is the most common explanation.
4. 4Once a technical mistake is excluded, mount a deliberate search for the unknown factor causing the deviation; this search is where the discovery lives ('It is when experiments go wrong that we find things out').
5. 5Interpret the clue using knowledge without fixed ideas, imagination, and scientific taste — hold the observation up against your body of knowledge while resisting premature explanation.
6. 6Keep a standing habit of contemplating all unexplained observations rather than discarding them, so that a fleeting anomaly is arrested before it slips away.
7. 7Strip out the word 'luck' — treat the event as chance plus preparation, never as a mystical influence; the observation only becomes a discovery through the trained mind acting on it.

Darwin, as described by his son, illustrates the operational stance. Everybody notices a striking, frequent exception as a fact, but Darwin 'had a special instinct for arresting an exception.' A point apparently slight and unconnected with his present work is passed over by many a man almost unconsciously with some half-considered explanation that is in fact no explanation. Darwin instead seized on just these things to make a start from. The chance opportunity here is not a dramatic accident but an ordinary anomaly most workers dismiss; the difference is the unremitting sensitivity that stops and questions it. The lesson is that the opportunity was equally available to others — only Darwin's habit of arresting the exception converted it into a starting point.

• If a result fails to reproduce with known factors unchanged, then first check for technical error; only after excluding it, treat the variation as a lead to an unrecognized factor.
• If an observation is slight, unexpected, and unconnected to your current problem, then arrest it and record it instead of explaining it away with a half-considered account.
• If you find yourself reaching for the word 'luck' or 'accident', then reframe: the event is chance, and the discovery depends on your preparation to notice and act.
• If an anomaly seems promising, then investigate it now; if it is unexplained but not yet promising, then log it into a standing set of observations you contemplate over time.
• If you are between active experiments for long stretches, then increase bench time or introduce a novel procedure, because opportunities come more frequently to active bench workers and those who dabble in new methods.

• Passing over an exception with a half-considered explanation — the tell is a quick, unexamined 'that's probably just X' that closes the question rather than opening it.
• Discarding non-reproducible results as failures instead of interrogating them — the tell is calling the work off as 'sloppy' before searching for the hidden factor.
• Assuming the anomaly is a real discovery before ruling out technical error — the tell is skipping the mundane check because the anomaly is exciting.
• Treating unexpected findings as 'luck' — the tell is mystical or fatalistic language that removes the observer's responsibility to prepare and notice.
• Being an inactive or purely theoretical worker who rarely touches the bench or tries new procedures — the tell is that opportunities almost never seem to arise, because exposure frequency is too low.

• ▸Observation: state the unexpected happening exactly as seen, no explanation attached.
• ▸Context: what experiment/procedure was running, and whether it was a novel technique.
• ▸Reproducibility: did known factors change? Yes/No — note any variation.
• ▸Technical-error check: ruled out? how? (most common explanation, so verify first).
• ▸Half-considered explanation to resist: write the easy dismissal you are tempted by, then set it aside.
• ▸Possible significance: what known-but-not-fixed idea could this connect to?
• ▸Promising now? Yes = investigate immediately / No = keep under contemplation.
• ▸Action taken / next probe: the specific step to hunt the unknown factor.
• ▸Review date: when to revisit unexplained observations still open.

1. 1Expose yourself to opportunity: spend long hours at the bench and try novel procedures rather than confining yourself to conventional ones, since significant clues are infrequent and favour the working, unconventional mind.
2. 2Stay alert for the unexpected while watching for the expected—treat any sterile culture, contaminant, side reaction, or anomalous result as a possible clue rather than a nuisance to discard.
3. 3When an anomaly appears, pause before discarding everything; instead run a small deliberate test that exploits the anomaly (as Pasteur re-inoculated the same old fowls before starting afresh).
4. 4Interpret the clue by relating it to the whole body of what you already know—give the observation significance it does not intrinsically possess by connecting it to other knowledge and possible implications.
5. 5Judge the evidence on its merits independently of prevailing conceptions; ask whether the observation is genuinely new using a good fund of knowledge across more than your own narrow field.
6. 6Triage the many side issues an alert mind encounters: accept you cannot follow all, and use scientific taste to pick the few worth investigating.
7. 7Follow the promising clue up by breaking it into crucial questions and designing an experiment that is consistent with one hypothesis and inconsistent with another.
8. 8Do not reject a line merely because others thought of or tried it without result—many classic discoveries were anticipated but not developed until the right worker pursued them fully.

Pasteur's fowl cholera cultures went sterile over the vacation—an unexpected obstacle he was about to discard entirely. Most sub-cultures failed to grow and injected birds were unaffected, so starting afresh seemed the sensible course. Instead he had the inspiration of re-inoculating those same fowls with a fresh, virulent culture. To general surprise nearly all these previously-exposed fowls withstood the inoculation, while fresh fowls succumbed after the usual incubation. The clue—weakened old cultures failing to kill—had no intrinsic significance until Pasteur related the survival pattern to the idea of protection. This yielded the principle of immunisation with attenuated pathogens. The lesson: the accident provided the opportunity, but the prepared mind that ran the confirming re-inoculation created the discovery.

• If an experiment throws up an unexpected result, then run a cheap confirming test that exploits it before discarding the material and starting over.
• If a clue seems trivial in isolation, then evaluate it by relating it to all you already know—significance is conferred by connection, not by the observation itself.
• If a promising idea has already been tried by others without success, then do not automatically reject it; the right development may not yet have been made.
• If you face many interesting side issues, then follow only the few your judgment rates promising and concentrate effort once something significant appears.
• If your judgment aligns suspiciously well with the prevailing conception, then suspect conditioned thinking and weigh the evidence independently on its merits.

• Noticing the clue but failing to appreciate its significance—the tell: like the physicist before Röntgen you feel merely annoyed by the anomaly, or like those who saw moulds inhibit staphylococci before Fleming, you note it and move on.
• Discarding anomalous material too fast—the tell: you are 'about to start afresh' the moment results don't fit expectation.
• Restricting activity to conventional procedures—the tell: your bench work never exposes you to fortunate accidents because you never try novel methods.
• Interpreting the clue prematurely and wrongly by not first studying the thing caused—the tell: Noguchi-style, you report a plausible cause (spirochaete for yellow fever) that later delays the whole field.
• Failing to transfer the clue across fields—the tell: you see a discovery announced elsewhere but assume it has no bearing on your own problem, so applications lie dormant for years.

• ▸Observation: state the exact chance/unexpected result as seen
• ▸Context: what were you actually trying to do when it appeared
• ▸Expected vs. actual: what should have happened and what did
• ▸Existing knowledge links: which known facts/principles could this connect to
• ▸Novelty check: is this new, or reported before—and was it ever properly developed
• ▸Possible significance: one hypothesis this clue supports and one it contradicts
• ▸Confirming test: the cheapest crucial experiment to run before discarding anything
• ▸Transfer scan: which other fields or problems this clue might apply to
• ▸Decision: pursue now / park / drop—and the reason

1. 1Classify your situation before deciding to persist: (a) stubborn adherence to an idea untenable in face of contrary evidence, or (b) persevering with a hypothesis hard to demonstrate but against which there is no direct evidence. Only category (b) earns perseverance.
2. 2Treat negative results as weak evidence for abandonment, as Bernard observed they mean very little; distinguish 'no support yet' from 'direct contradiction.'
3. 3When facing indifference and opposition, expect the discovery to 'run the gauntlet of scepticism and often resistance'; prepare to back your conviction rather than drop the line at the first cold reception.
4. 4Back your own hypotheses, not borrowed ones — you work much better pursuing your own idea because you hold the strong desire to confirm it that drives you to try all variations of conditions.
5. 5Guard perseverance against parental affection: subordinate the idea to the facts, and run a rigorous test for error before reporting, so tenacity does not become wishful thinking.
6. 6Set an exit condition: even within category (b), if no progress is being made, shelve the project — sometimes for years — and take it up again when fresh knowledge or a new approach appears.
7. 7Detect and correct errors fast rather than avoiding risk; the best single measure of mental skill is the speed with which errors are detected and thrown out, so venture boldly but recant publicly when wrong.

Jenner's recognition of vaccination shows perseverance and courage together. The discovery took him thirty years, and others with the same opportunity failed to make it, showing how difficult it was. The idea was revolutionary: animals were regarded with repugnance, so infecting a human with a disease of animals created utmost disgust, and dire results were prophesied — 'cow-mania' and 'ox-faced children,' one of which was actually exhibited. The discovery required little erudition; it mainly devolved on having boldness and independence of mind to accept a revolutionary idea plus imagination to realise its potentialities. Jenner also faced practical frustrations — cows had various teat sores, some affecting milkers without giving immunity, hard to distinguish even for present-day specialists, and cow-pox did not reliably confer immunity in the cow itself. He persisted through both the opposition of others and these repeated technical difficulties before performing the crucial experiments.

• If facts directly contradict the hypothesis or force unduly complicated, improbable subsidiary hypotheses, then discard it with as few regrets as possible.
• If the hypothesis is merely hard to demonstrate but has no direct evidence against it, then persevere — like Faraday persisting through repeated failures before producing current from a magnet.
• If you are persevering in category (b) but no progress is being made, then shelve the project temporarily and return when techniques, related knowledge, or a new approach are available.
• If the idea is not your own and you feel no strong desire to confirm it, then recognise you will likely abandon it prematurely — either make it your own or hand it to its originator.
• If the obstacle is indifference and opposition rather than contrary evidence, then treat it as a hurdle to fight through, not a signal to abandon.

• Clinging to a sterile idea after facts warrant its discard — Zinsser's image of hens sitting on boiled eggs; the tell is explaining away observations that don't fit the pattern you've woven.
• Confusing stubbornness with perseverance: persisting against direct contrary evidence rather than merely against difficulty of demonstration; the tell is that new facts keep contradicting you, not just failing to confirm.
• Letting parental affection turn tenacity into wishful thinking, so observations and interpretation bend to protect the brain-child; the tell is reluctance to run a rigorous error test before reporting.
• Abandoning a hypothesis that is not your own after one or two attempts because you lack the desire to vary conditions and give it a thorough trial; the tell is you never think out all possible ways to change the experiment.
• Excessive caution that avoids all risk of going astray — 'panic of error is the death of progress'; the tell is you make neither errors nor discoveries.

• ▸Hypothesis (one sentence): state the idea you are backing.
• ▸Whose idea: mark yours or borrowed — borrowed ideas need re-owning or handing back.
• ▸Evidence against: list any DIRECT contrary evidence versus mere failure to confirm.
• ▸Category: (a) untenable vs contrary evidence, or (b) hard to demonstrate, no evidence against.
• ▸Nature of obstacle: technical difficulty / indifference / opposition / missing related knowledge.
• ▸Progress check: is any progress being made now? yes / no.
• ▸Exit condition: if no progress by [date/milestone], shelve until [new knowledge or approach].
• ▸Error test done: rigorous check for error completed before any claim? yes / no.
• ▸Decision: persevere / modify / shelve temporarily / discard — with reason.

1. 1Separate the experimental result from its interpretation before defending or attacking any claim. When you meet a new idea, first ask whether you are reacting to the evidence or to the disturbance the idea causes; the passages call the automatic response the 'attack-escape' reaction, where attack includes ridicule and escape includes putting the idea out of mind.
2. 2Identify the prevailing doctrine your discovery must oust. Following Butterfield, the hardest act is to re-arrange a familiar bundle of data and escape the reigning theory, so name explicitly the accepted notions (e.g. blood ebbing and flowing, two sorts of blood) that stand in the way.
3. 3Gauge whether the discovery arrives 'in time' or 'before its time.' If it grows out of and finds support in prevailing concepts, expect acceptance with only mild resistance; if it departs too far to be fitted into current knowledge, expect it to be ignored or opposed too strongly to overcome.
4. 4Distinguish psychological resistance (herd instinct, mental inertia) from institutional opposition (entrenched authority and vested interests). Innovations are opposed because they disturb people who hold high honours for past accomplishments; treat these as two separate hurdles requiring different responses.
5. 5Anticipate the delay mechanisms and prepare to persist: indifference, scepticism, lack of application until a supporting technique appears, and lack of follow-through by others. Plan for a fight measured in years, not weeks.
6. 6Trust your own observations even when they contradict expectation. Vesalius could hardly believe his eyes when structures did not match Galen; lesser men blamed the specimen or their own handiwork. Guard against ignoring results that fail to coincide with what you expected.
7. 7Read critically to keep your own mind from becoming the source of resistance. Following Bacon, read 'to weigh and consider,' not to confute or to believe, so that false or entrenched information does not become a barrier to new ideas.

Harvey's discovery of the circulation of the blood shows the full gauntlet. The prevailing beliefs — that blood ebbed and flowed in the vessels, that there were two sorts of blood, and that blood passed from one side of the heart to the other — had struck deep root and had to be ousted. His first dissatisfaction came from a concrete observation: the direction the valves faced in the veins. He needed courage even to announce how much blood he calculated the heart pumped, writing that his finding was 'of so novel and unheard-of character' that he feared 'mankind at large for my enemies,' because 'want and custom' and 'respect for antiquity' influence all men. His fears were justified: he was subjected to derision and abuse and his practice suffered badly. Only after a struggle of over twenty years did the circulation of the blood become generally accepted — the measure of how long institutional and psychological resistance can delay a sound discovery.

• If a discovery fits and finds support in prevailing concepts, then expect ready acceptance (unless made by an outsider) and lean on that alignment; if it departs too far to be fitted into current knowledge, then expect strong opposition and prepare for a long struggle.
• If you notice yourself arguing against an idea before it has been fully stated, then treat that as the antigen reaction Trotter described and suspend the response until you have weighed the evidence.
• If knowledge in the field is still growing, then advantage lies with the expert; if the field is worked out and needs a revolutionary approach, then expect the breakthrough from an outsider and expect experts to greet it with scepticism.
• If a result contradicts your expectation, then mistrust the expectation before mistrusting the observation — students and even Vesalius's contemporaries erred by blaming the specimen.
• If a discovery has no possible application yet, then expect indifference until a supporting technique arrives (Neufeld's typing waited until 1931; Landsteiner's blood groups waited for anticoagulants and transfusion).

• Turning a blind eye to previous ignorance — assuming there was no problem or that accepted notions settle it. The tell: you cannot state which prevailing doctrine your idea must displace.
• Mistaking rationalisation for reasoning. Instinctive resistance gets dressed up in secondary 'reasons' formed only to justify a pre-existing opinion. The tell: your objections crystallised before the idea was completely stated.
• Confusing a critical attitude with a sceptical one, or swinging to the opposite error of believing plausible ideas too readily. The tell: you are taking a side out of temperamental conservatism or progressiveness rather than judging the evidence.
• Disbelieving your own eyes and blaming the apparatus or specimen when a fact is unexpected. The tell: you discard or 'correct' results that fail to match your expectations.
• Failing to develop or follow through a discovery because others found similar stories that came to nothing — as happened between Fleming and Florey on penicillin. The tell: you abandon a finding for lack of encouragement rather than for lack of evidence.

• ▸Discovery stated plainly: write the finding, separating result from interpretation.
• ▸Prevailing doctrine it must oust: name the accepted belief(s) standing in the way.
• ▸Ripeness check: does it grow from current knowledge (in time) or depart too far (before its time)?
• ▸My own first reaction: attack (ridicule) or escape (put out of mind)? Note it honestly.
• ▸Evidence vs. expectation conflict: is any result being ignored because it surprised me?
• ▸Psychological resistance sources: herd conformity, mental inertia, respect for antiquity.
• ▸Institutional opposition sources: whose entrenched authority or vested interests are disturbed?
• ▸Application gap: is a supporting technique missing that would give the discovery importance?
• ▸Persistence plan: expected duration of struggle and next concrete step to press the case.

1. 1Study things caused before causes: define the disease or phenomenon by its manifestations and distinguish it from conditions it may be confused with, as Hughlings Jackson directs, before hunting the cause.
2. 2Do field observation and personal enquiry first (interview farmers, gamekeepers, lay observers), then separate their raw observations from their interpretations through tactful, searching questioning.
3. 3Marshal and correlate all data, then break the defined problem into several formulated questions and translate each into a planned experiment.
4. 4Design each experiment for reproducibility; establish each stage beyond reasonable doubt before advancing so the work cannot be condemned as sloppy.
5. 5Take great pains with essential technical details — check methods against known positive and negative specimens — since technical error is the most common source of misleading results.
6. 6When results vary despite unchanged known factors, welcome it: hunt the unrecognised factor, because 'it is when experiments go wrong that we find things out' — but first rule out technical mistake.
7. 7Hold hypotheses loosely: use them to indicate new experiments, but modify or discard as soon as contrary evidence appears, and guard your observations against bias toward the expectation.
8. 8If you judge a discovery important, back your judgment enough to run confirmatory trials elsewhere rather than abandon it on a single failed result.

Beveridge saw a demonstration of the Mules operation for preventing blowfly attack in sheep and recognised its significance immediately. He set up a trial involving thousands of sheep and, without waiting for his own results, persuaded colleagues in other regions to run parallel trials. When his results came in a year later, the sheep in his trial showed no benefit — an unexplained failure. But every other trial, and all subsequent ones, showed the operation conferred valuable protection. Because he had enough confidence in his judgment to prevail on colleagues to test it independently, the discovery was validated despite his own experiment going inexplicably wrong. Had he been more cautious and waited only on his own result, the finding would have been wrongly dismissed.

• If your results vary while known factors are constant, then suspect an unrecognised factor and investigate it — but check for technical error first.
• If lay informants offer evidence, then separate what they observed from how they interpret it, since undisciplined minds remember what supports their views.
• If contrary facts appear, then do not expect an idea to be abandoned until a better hypothesis replaces it — supply the replacement, don't just attack the old.
• If a single experiment fails but your judgment says the effect is real, then run independent replications elsewhere before discarding the finding.
• If you are entering a well-worn field, then limit conventional reading that conditions the mind into the same rut, so a new approach stays accessible.

• Rushing ill-planned, sketchy experiments out of impatience — the tell is results that cannot be reproduced and get condemned as 'sloppy'.
• Doing laboratory work with little relation to the real problem because field observation and enquiry were skipped — the tell is a lab result no farmer's experience corroborates.
• Confounding your own experiment through hidden contamination, like John Hunter inoculating himself with material containing both gonorrhoea and syphilis and cementing a false unified-disease belief for years.
• Becoming too attached to a hypothesis so observations get biased toward the expected — the tell is Mendel-like data that, examined coldly, is 'not entirely objective'.
• Studying causes before defining the thing caused, like Noguchi naming a spirochaete the cause of yellow fever, which delayed the whole investigation.

• ▸Thing caused (define it): list manifestations and what distinguishes it from confusable conditions.
• ▸Field enquiry log: who observed what — record raw observations only.
• ▸Interpretation separated out: note where informants coloured evidence to fit notions.
• ▸Formulated questions: break the defined problem into specific testable questions.
• ▸Reproducibility check: how will this experiment be repeated and by whom.
• ▸Technique controls: known positive and negative specimens used to verify method.
• ▸Hypothesis and disconfirmer: state it, and the evidence that would make you discard it.
• ▸Anomaly note: unexplained variation — is it technical error or a new factor?
• ▸Confidence action: if result fails but judgment holds, where do you replicate?

1. 1Separate the precursor from the developer: when you claim a discovery matters, ask who first suggested it versus who forced it on a reluctant world. Credit and impact attach to the one who develops the idea to fruition, as with Columbus, Pasteur, and Lister who were not first but who fully developed the ideas.
2. 2Do not stop at the initial disclosure — follow up and exploit it. After obtaining new knowledge, use it to uncover something further, because failure to follow up delays general acceptance more than any single obstacle.
3. 3Find or build the application. A finding without an applied outcome stalls; penicillin sat as a crude 1929 description until Florey supplied the missing step of developing a therapeutic agent.
4. 4Supply the missing explanatory mechanism, because acceptance often waits on it. Semmelweis' teaching was rejected partly because there was no satisfactory explanation for disinfecting hands until bacteria were shown to cause disease.
5. 5Manage the human relations of the discovery. Anticipate that innovations threaten entrenched authority and vested interests, and use tact and conciliation to lower resistance rather than logic alone.
6. 6Distinguish critical judgment from mere scepticism. Aim for honest, objective judgment of the evidence, suspending judgment where evidence is incomplete, rather than defending either conservatism or progressiveness.
7. 7Accept a lag as normal and use it. Expect years or decades between disclosure and recognition; plan for others to solve the problem independently and for credit to settle only when the world is forced to see.

Fleming described crude preparations of penicillin in 1929 but dropped the work within a few years without producing a therapeutic agent. He got no encouragement or assistance because others knew of many similar stories that had come to nothing — the failure was in following up and in the absence of an application, not in the original observation. Years later Florey took the work up from where Fleming left off and developed penicillin into a usable therapeutic agent, at which point its impact became undeniable. The same pattern recurs with cod-liver oil: Steinhaeuser showed in 1840 that it cured rickets, yet the fact remained an unproved opinion for eighty years because no one exploited it. Theobald Smith's 1903 work on motile and non-motile bacillary variants passed almost unnoticed and was forgotten until Weil and Felix rediscovered it in 1917, after which it became regarded as one of the fundamental facts of immunological reactions. In each case the discovery achieved impact only when someone supplied the follow-up, the application, or the rediscovery that the originator did not.

• If a claim is 'not entirely original,' then trace its precursor ideas and assign impact to whoever developed it to fruition, not to whoever first suggested it.
• If a finding has been disclosed but not followed up, then treat the follow-up and its application as the real deliverable — Fleming's gap versus Florey's development.
• If a correct discovery is being rejected, then check whether the explanatory mechanism is missing (Semmelweis before germ theory) before blaming the audience.
• If opponents argue 'it does not solve the whole problem, so it is useless,' then reject that all-or-nothing attitude and adopt the development as useful in the absence of anything better.
• If the discoverer is combative or tactless, then expect delayed acceptance; Harvey's diplomacy succeeded where Semmelweis' lack of tact failed.

• Abandoning a finding for lack of encouragement — the tell is that everyone cites 'similar stories that came to nothing,' as happened to Fleming before Florey.
• Treating the initial result as the endpoint — the tell is an important fact that remains an unproved opinion for years, like cod-liver oil for rickets across eighty years.
• Relying on logic to compel acceptance — the tell is persecution of the discoverer and reasoned resistance to obvious facts, since logic was built to win debates, not to advance science.
• Too-ready acceptance of plausible ideas on slender evidence — the inexperienced scientist's error; the tell is conviction of causation without controls, the post hoc, ergo propter hoc trap behind faith in worthless medicines.
• Ignoring lay or untrained observation because it conflicts with expert preconception — the tell is a dismissed report later found true, like the gamekeeper's claim about ferrets and canine distemper, or Italian peasants' two-thousand-year belief about mosquitoes and malaria.

• ▸Discovery / claim: state the finding in one line
• ▸Precursor vs developer: name who first suggested it and who developed it to fruition
• ▸Follow-up status: what further question did the initial finding open, and was it pursued
• ▸Application: name the practical or theoretical use that carries the impact
• ▸Missing mechanism: what explanation, if absent, is blocking acceptance
• ▸Source of resistance: entrenched authority, vested interest, or preconceived idea
• ▸Discoverer diplomacy: tactful (Harvey) or combative (Semmelweis) — note the risk
• ▸Evidence standard: is it proved and reproducible, or accepted on slender evidence
• ▸Expected lag: estimate time to recognition and note if others may solve it independently

1. 1Ask an experienced worker in your field which twenty to forty periodicals matter most, then commit to at least glancing through their article titles regularly, the way you read a newspaper.
2. 2Skim-read broadly for coverage: read quickly through each article to get a picture of the whole, but stop skimming the moment you hit closely reasoned or condensed writing, which demands close study.
3. 3Keep a card index with brief abstracts of articles of special interest; write the abstract yourself so the act of condensing impresses the salient features into memory.
4. 4After the first skim, return to the parts whose full significance is now apparent, re-read them, and make notes — this two-pass method separates coverage from depth.
5. 5Deliberately build foundations: get a thorough grounding in the fundamental sciences, because in science the adult mind can build only as high as the foundations laid in youth will support.
6. 6Study for perspective, not accumulation — prioritise a clear grasp of general principles (held loosely, not as fixed laws) over memorising detail retrievable from reference books.
7. 7Widen your base beyond your speciality: teach if you can (it forces you to keep abreast of a wider field), take a course in biometrics, read the lives of great scientists, and learn some history of science as a corrective to over-specialisation.

A biology graduate entering infectious-disease research first asks a senior worker which journals to follow, and settles on roughly thirty periodicals to scan weekly. Rather than learn German — once standard, but now less useful given the collapse of German biological output and the rise of English as the international scientific language — he spends his first year on a course in biometrics. Each week he skims titles, reads fully only the few articles bearing on his problem, and skips skimming on any densely reasoned paper he intends to study closely. For the articles that matter, he reads through once for the whole picture, then returns to re-read the significant sections and writes a brief abstract onto a card index. Over time this store of assessed knowledge lets him quickly judge new work and grasp its significant findings — building the background from which useful combinations of ideas can arise.

• If a piece is closely reasoned, condensed, or something you intend to study carefully, then do not skim it — read it slowly and completely.
• If you are choosing between loading detailed technical information and understanding general principles, then favour the principles — details live in reference books and card indexes.
• If you lack a special reason to learn German, then spend the effort on biometrics instead, given the shift of scientific publishing toward English.
• If you find an article of special interest, then write your own brief abstract for the card index rather than relying on published abstracting journals, which lag behind the originals.
• If you are over-specialising, then correct it by teaching, reading scientific biographies, or studying the history of science to broaden outlook.

• Trying to keep abreast of the whole literature — with ~2 million articles a year, this is impossible; the tell is drowning in reading while covering none of it properly.
• Seeing only the trees: loading the mind with detail retrievable from reference books while missing the general principles; the tell is technical fluency with no perspective on significance.
• Treating general principles as fixed laws rather than working generalisations; the tell is rigidity and inability to reassess new work.
• Skimming material that demands close study, so significant findings are missed; the tell is having 'covered' a paper you cannot actually assess.
• Skimming without an adequate background of knowledge, so you cannot quickly assess new work or spot what is significant; the tell is fast reading that yields no usable judgment.

• ▸Core journals (20-40): list titles a senior worker recommended for your field
• ▸Weekly skim pass: date and which titles/journals you scanned
• ▸Close-study queue: articles too reasoned to skim, flagged for full reading
• ▸Card-index abstract: your own brief summary of each article of special interest
• ▸Salient findings: the significant points to re-read after the first whole-picture pass
• ▸Principles noted: general principles extracted, marked as provisional not fixed
• ▸Foundation gap: fundamental-science or biometrics area to strengthen
• ▸Broadening activity: teaching, a scientist's biography, or history-of-science reading done this period

1. 1Notice, don't skip, the phenomena you meet that have no satisfactory explanation — the scientist's curiosity is directed at exactly these unexplained things or relationships, so log them the moment you notice them.
2. 2Convert the noticing into an explicit question: ask 'how does this process work?' or 'why does it take this form?' The question itself is the stimulus that produces ideas; a difficulty you can't phrase as a question won't drive thought.
3. 3Prefer 'why' when a cause or purpose is plausible — in biology it is justified because all events have causes and structures usually serve a survival function — and use 'how' to probe the mechanism. Let the question point at an imagined cause you can test.
4. 4Design the smallest concrete manipulation that would answer the question. Ask 'what would happen if...' and pick an intervention whose result would distinguish between explanations (as Hunter did with the carotid ligature).
5. 5Run it, then check your assumptions when the result surprises you — verify the manipulation still held before revising your explanation; the anomaly is often the discovery.
6. 6Guard curiosity against being blunted: keep transferring it to intellectual interests, seek the underlying principle tying masses of unrelated data into an orderly whole, and treat each answer as a higher vantage that opens a wider field of new questions.
7. 7Protect the drive socially — share problems at lunch, tea, and seminars; enthusiasm is infectious and is the best safeguard against the doldrums when curiosity flags.

In Richmond Park, John Hunter saw a deer with growing antlers and simply wondered what would happen if he shut off the blood supply on one side of the head — curiosity noticing an unexplained relationship. He turned the wonder into a manipulation: he tied the external carotid artery on one side, and the corresponding antler lost its warmth and stopped growing. Rather than stopping there, he kept watching, and after a while the antler grew warm again and resumed growth. Instead of assuming his ligature had failed, he checked it — it still held — and found that neighbouring arteries had enlarged to carry adequate blood, thereby discovering collateral circulation and its capacity to increase. He then extended the finding to a problem no one had dared touch: aneurism had never been treated by ligation for fear of gangrene, but Hunter now saw the possibility and tried ligation on a popliteal aneurism, establishing the Hunterian operation. His insatiable curiosity was the driving force — he even paid a surgeon's expenses to go observe whales.

• If you notice something with no satisfactory explanation, then log it and phrase it as a 'why' or 'how' question rather than moving on.
• If a structure or reaction plausibly serves a function or has a cause, then asking 'why' is legitimate and productive; if you only need the mechanism, ask 'how'.
• If an experimental result reverses or surprises you, then verify your manipulation is still intact before revising the explanation — the anomaly may be the finding.
• If you've answered your question, then treat the answer as a higher vantage and immediately ask what wider field it now reveals — there is no satisfying curiosity.
• If your curiosity feels blunted, then transfer it to fresh intellectual interests or seek an ordering principle in unrelated data, and seek stimulating discussion to reignite it.

• Coasting on a working knowledge of the environment so curiosity goes blunt — the tell is that you acquire new knowledge only by being told or reading, never by asking why something is so.
• Failing to formulate the difficulty as a question — the tell is that no ideas are produced because there's no stimulus; people with no curiosity seldom get this stimulus.
• Refusing to ask 'why' out of teleology-phobia and thereby missing the imaginative prompt about cause or purpose — the tell is treating the ban on supernatural purpose as a ban on asking about function at all (Von Bruecke: the biologist lives with teleology but is ashamed to be seen with her).
• Assuming failure when an experiment reverses instead of checking the apparatus — the tell is discarding a result Hunter would have investigated (he confirmed the ligature held and found collateral circulation).
• Treating each answer as an endpoint rather than a wider vantage — the tell is that discovery stops after the first finding instead of extending it, as Hunter did from collateral circulation to the aneurism operation.

• ▸Unexplained observation: the thing or relationship you noticed that has no satisfactory explanation
• ▸Question form: state it as 'how does X work?' or 'why does X take this form?'
• ▸Why/How justification: name the plausible cause or function that legitimizes the question
• ▸Imagined cause: your candidate explanation to be tested
• ▸Minimal manipulation: the smallest 'what if I change this?' intervention that would distinguish explanations
• ▸Predicted vs actual result: what you expected, what happened, and any surprise
• ▸Assumption check: confirm the manipulation held before revising (did the ligature still hold?)
• ▸Next vantage: the wider question this answer now opens up
• ▸Discussion note: whom you'll talk it over with to keep enthusiasm alive

1. 1When reading a report, physically separate two columns: (1) the experimental results as reported, (2) the author's interpretation of them. Believe neither on sight; weigh and consider each per Bacon's rule — read not to contradict nor to take for granted.
2. 2As you read, correlate every claim against your own knowledge and experience, actively hunting for significant analogies, generalisations, gaps in present knowledge, and inconsistencies between different writers or between reports and your local observations.
3. 3Use reading as a stimulus to thinking rather than capital to accumulate; pair it with active research so freshness of outlook is not lost. Ask what problem each fact might attack, not merely file the fact.
4. 4Hold your own hypotheses to a harder standard than others' — remember the originator identifies with his idea and grows devoted to it. Deliberately design a crucial experiment that is consistent with one hypothesis and inconsistent with a rival.
5. 5When results contradict your hypothesis, resist patching it with ad hoc subsidiary hypotheses; count the patches. When the edifice grows ridiculously overburdened, break it down and build a synthesis that fits all facts now available.
6. 6Protect assessment from your own bias: where results are subjective (clinical severity, histological change), arrange that the judge does not know which group each case belongs to, since you cannot be sure your judgment is not subconsciously biased.
7. 7Reserve strict demand for convincing evidence for the proof stage; during the search stage keep the mind malleable and open to the unexpected and revolutionary — the boldness and independence Jenner needed to accept a revolutionary idea.

Noguchi isolated a spirochaete from cases of leptospiral jaundice and reported it as the cause of yellow fever — an understandable mistake that delayed yellow fever investigations for years. The failure was one of the critical-open balance: he accepted his own finding as cause without holding it to the harder standard rivals would apply, and did not distinguish a correlated organism from the true agent. A reader with the right attitude, reviewing Noguchi's report, would have split result (spirochaete isolated) from interpretation (it is the cause), then looked for inconsistencies with other writers' reports. Contrast Jenner: for thirty years he weighed the popular belief about cowpox against evidence, distinguishing the several sores on cows' teats — only some of which gave immunity — before performing the crucial experiments. His discovery required boldness to accept a revolutionary idea plus the discipline to test it, not erudition. The pairing shows both failure modes: Noguchi over-believed his own experiment; the many who let the cowpox opportunity pass over decades over-deferred to authority and repugnance.

• If the passage states a result, THEN treat it as provisional data; if it states what the result means, THEN treat it as a separate hypothesis to be tested independently.
• If you are the originator of the hypothesis, THEN distrust it more, not less — you feel confident but should recall the many little things that could have gone wrong; if you are the critic, THEN state objections gently, since scorn discourages the originator.
• If contrary facts appear and you must add a subsidiary hypothesis to save your main one, THEN log it; if the additions become numerous and strained, THEN abandon the edifice and rebuild.
• If results can be judged subjectively, THEN blind the judge to group membership; if they can be measured objectively, THEN prefer the measurement.
• If the field is worked-out and static, THEN weight a revolutionary outsider's approach and expect experts' scepticism to signal that existing knowledge is a handicap; if knowledge is still growing, THEN weight the expert.

• Believing everything you read — the young scientist's most common mistake; the tell is that he cannot state which part of a paper is measured result and which is the author's interpretation.
• Accumulating information as capital investment without thinking; the tell is a full notebook but no hypotheses, analogies, or noticed gaps emerging from the reading.
• Becoming devoted to your own hypothesis and defending it with ever more ad hoc additions; the tell is a growing list of special-case exceptions that never triggers a rebuild.
• Over-trusting experiments and averages; the tell is reporting results only as averages with no frequency distribution, or joining sparse graph points with lines that hide an unsuspected rise and fall.
• Excessive scepticism toward revolutionary ideas or over-deference to authority; the tell is greeting a new approach with reflexive dismissal, or the language-trained habit of accepting authority instead of seeking evidence.

• ▸Reported result (verbatim) — copy only what was observed/measured
• ▸Author's interpretation — copy the causal or explanatory claim, kept separate
• ▸Weigh & consider — your judgment: which is supported, which is a leap?
• ▸Analogies / generalisations — related problems or patterns this connects to
• ▸Gaps & inconsistencies — missing knowledge or conflicts between writers
• ▸Rival hypotheses — at least two, and the crucial test distinguishing them
• ▸Bias check — is assessment subjective? if so, how will the judge be blinded?
• ▸Ad hoc patch count — subsidiary hypotheses added; rebuild if strained
• ▸Field status — growing vs worked-out; whom to weight (expert vs outsider)

1. 1Digest the problem thoroughly before formulating any experiment: collect firsthand field data, marshal and correlate it, then break the problem into specific crucial questions. Nicolle and Pasteur did few, simple experiments precisely because they invested long hours of intellectual incubation first.
2. 2Generate as many hypotheses as possible to answer each question — 'intelligent guesses,' held as tools, not conclusions. Do not stop at the first plausible one.
3. 3Rank hypotheses and questions by importance, then devise experiments to test the likeliest hypotheses bearing on the most crucial questions first.
4. 4Design a crucial experiment where possible: one that gives a result consistent with one hypothesis and inconsistent with another, so the outcome discriminates rather than merely confirms.
5. 5Run the experiment under known conditions with as many extraneous influences eliminated as possible; use a controlled experiment with similar groups identical except for the variable under test.
6. 6Check the hypothesis against the world of observable facts. When a result contradicts the hypothesis, modify or abandon it — do not defend it. Hold each one tentatively.
7. 7Guard the attitude: retain the zeal of the fanatic for pursuing the idea, but discipline it with objective judgment of results and active exposure to criticism from others.

Consider the discovery that ferrets are susceptible to canine distemper, which arose from a gamekeeper's assertion. The scientists at first did not take his statement seriously — it ran against their expectations. But rather than dismissing it permanently on the strength of their prior view, they later decided to see if there was anything in it. Treating the gamekeeper's claim as a hypothesis worth testing, not as folk error to be filtered out, led to an important finding. The lesson is that a hypothesis is a tool to suggest a new observation or experiment; even an unwelcome one earns a test. Had they stayed attached to their assumption that a layman could not have observed something real, the discovery would have been lost.

• If a result is consistent with your hypothesis AND inconsistent with a rival hypothesis, then treat that experiment as crucial and prioritise it.
• If field or observational data contradict your working hypothesis, then modify or discard the hypothesis rather than the data.
• If you find yourself defending a hypothesis against contrary facts because it is your 'brain-child,' then deliberately invite outside criticism and consider abandoning it.
• If one hypothesis becomes significant early, then concentrate efforts there rather than dispersing across too wide a front.
• If an untrained observer (e.g., a farmer or gamekeeper) reports something contrary to expectation, then separate their observation from their interpretation and consider testing it before rejecting.

• Excessive attachment to one's own hypothesis — the tell is that you interpret contrary results as experimental error while accepting confirming results uncritically.
• Rushing into 'frantic and often ill-conceived experimental activities' (Zinsser's 'ant-like agitation') without long preliminary thought — the tell is high output of 'what-of-it' literature and few conclusive experiments.
• Considering only one hypothesis instead of as many as possible — the tell is you never designed a crucial experiment because you had nothing to contrast against.
• Letting existing knowledge bias the mind so a genuine anomaly is dismissed — the tell is reflexive scepticism toward the unexpected report ('not taken seriously at first').
• Confusing observation with interpretation, in yourself or your informants — the tell is remembering events that support your view and forgetting those that do not.

• ▸Crucial question: state the single sharpest question this hypothesis addresses.
• ▸Hypotheses considered: list all plausible answers generated, not just the favoured one.
• ▸Chosen hypothesis: the one to test first, with why it is most crucial/likely.
• ▸Rival hypothesis: the competing explanation the experiment must rule out.
• ▸Crucial test design: the experiment giving a result consistent with one and inconsistent with the other.
• ▸Controls: the similar group(s) identical except for the variable, extraneous influences eliminated.
• ▸Falsifying result: state in advance what observation would force you to modify or abandon this hypothesis.
• ▸Attachment check: name who will critique the result and how you will act if facts contradict you.

1. 1Load the mind first: accumulate relevant but loosely connected data across your field and beyond, since original combinations require raw material already present to associate. A clarifying idea crystallizes only when a nucleus of correct data is already in solution.
2. 2Read widely but calibrate depth: study your immediate problem closely, skim summaries and reviews outside it, and correlate everything read with your own knowledge, looking for significant analogies and generalizations rather than storing detail.
3. 3Treat the idea-generating step as involuntary. Follow Dewey's phases: become aware of the difficulty, let a suggested solution spring up, then bring reason in to accept or reject it. If reason rejects it, revert and repeat — do not force the flash.
4. 4Deliberately seek links between things whose connection was not previously suspected; originality often consists in this linking, as with survival of the fittest reaching Darwin and Wallace independently.
5. 5Judge each side-issue clue for whether to pursue it, since you cannot follow all of them; distinguishing the promising few is the essence of the art.
6. 6Guard against your own stored knowledge: recognize that a mass of information insufficient for the problem, or containing false items, actively blocks new ideas, so hold general principles as flexible, not fixed laws.
7. 7Have enough confidence in judgment to act on a fired imagination even before confirming results, while keeping the supposition provisional and ready to be modified.

Beveridge watched a demonstration of the Mules operation for preventing blowfly attack in sheep and his imagination was immediately fired by its potentialities — an intuitive grasp of significance preceding proof. Acting on that judgment rather than waiting, he set up a trial involving thousands of sheep and persuaded colleagues elsewhere to run their own experiments. When his own trial returned about a year later, it showed no benefit, and no satisfactory explanation for the failure was ever found. Yet every other trial confirmed the operation gave valuable protection. Had he been cautious and waited for his own misleading result, adoption would probably have been retarded for years. The episode shows imagination supplying the initiative and confidence carrying it, while a single anomalous experiment did not overturn a judgment grounded in wider evidence — note also that Mules himself, with no scientific training, made the original discovery where trained scientists had failed.

• If your field is still growing or the problem is new, then favor the expert with a deep fund of knowledge; if the field is worked out and stalled, then expect the revolutionary approach from an outsider free of fixed ideas.
• If a promising idea has already been thought of or tried without result, then do not reject it on that ground alone — many classic discoveries were anticipated but not developed until the right person came (Jenner, Harvey, Darwin).
• If your imagination is fired by an observation's potential, then act to set experiments in motion, but treat the supposition as provisional and hold it without believing it.
• If you cannot conjure an original idea despite ample data, then suspect the stored information is insufficient or partly false and step back to general principles rather than piling on more detail.
• If reason rejects a flashed idea, then revert to awareness of the difficulty and let the process repeat rather than defending the idea.

• Believing everything you read and failing to separate reported results from the author's interpretation — the tell is that you cite conclusions without checking whether the experiments actually support them.
• Loading the mind with detailed technical information while missing general principles — the tell is seeing only the trees, unable to see the wood, and creative thinking drying up.
• Letting established knowledge bias you: 'it is that which we do know which is the great hindrance to our learning.' The tell is instant scepticism toward a revolutionary idea, which itself signals the field's knowledge is inadequate.
• Forcing the idea by voluntary effort — the tell is treating the flash of insight as something you do rather than something that happens to you, so you grind without producing.
• Over-cautiously waiting for a single experiment before acting, when it may be misleading — the tell is retarding adoption of a sound discovery because one anomalous trial (like Beveridge's Mules sheep) contradicted the weight of evidence.

• ▸Problem/difficulty: state the specific stimulus you are stuck on in one sentence.
• ▸Relevant data on hand: list the loosely connected facts already in your mind for this problem.
• ▸Cross-field reading: note two areas outside your specialty to skim for analogies.
• ▸Suspected links: write pairs of ideas whose connection was not previously suspected.
• ▸Flashed idea: record the suggested solution exactly as it occurred, before judging it.
• ▸Reason check: mark accept / reject / revert and the evidence for that decision.
• ▸Prior-attempt status: has this been tried before, and is that a real reason to drop it?
• ▸Confidence-to-act: is imagination fired enough to set experiments in motion now, yes/no and why.
• ▸Held-as-provisional flag: confirm the supposition is used without being believed and name the evidence that would modify it.