# Survival of Fitness — War of the Worlds, Part I
## Historical research packet and MRI-003 specification

**Prepared September 20, 2026. Version 1.0. Research, not an essay draft.**

**Question:** How did a period of extraordinary scientific and cultural activity move from describing human variation toward claims about what humanity should become?

## Read this first: what this packet establishes

The surviving publication distinguishes **a meeting from an edited discussion record**. George Bernard Shaw is printed under **Written Communications**, not the spoken Discussion. Alice Drysdale Vickery is listed in the Discussion as **“Mrs. Dr. Drysdale Vickery.”** Pearson is the chair. Earlier descriptions of all these people speaking together in one room must be corrected. Written contribution does not prove absence; it simply does not establish physical attendance. [S01–S03]

The numerical work also has two layers. Galton’s ten normal-probability classes reproduce to their published rounded counts. The subsequent comparison with Booth uses explicit approximation and regrouping. The descendant table is a theoretical construction with stipulated assumptions, not an observed intergenerational dataset of civic worth. This packet contains transcriptions and executed arithmetic checks; it does **not** claim full reproduction of the descendant model. [S04–S05; `results/arithmetic_checks.json`]

**Completed:** primary discussion text review; original-volume contents check; distinction between recorded oral and written contributions; primary-source-led timeline; eight focused cast profiles plus additional voices; a complete 19-name publication roster; codebook and 25 illustrative seed propositions; scan transcription of Tables I–III; executable numerical checks; Wells visual specification; source ledger; factual vetoes; narrative-bearing findings.

**Not completed:** a fully double-coded discussion corpus; independent second-person transcription review; reconstruction of the original Booth figures upstream of Galton; exact replication of every Table III cell; complete facsimile verification of the descent figure and Table IV; a public interactive. Original PDFs and page images were inspected through web tools but could not be downloaded into the execution environment. No original scans are bundled.

### Evidence labels

- **F — documentary fact:** the document, date, printed words, section placement, or bibliographic artifact can be checked.
- **C — contemporary claim:** what a participant asserts, recommends, disputes, or assumes. Its appearance in the record does not establish its truth.
- **H — later historical account:** a custodian’s biography or scholarship used for context or identity; explicitly attributed.
- **I — proposed interpretation:** a question or reading supplied by this project, not a quotation or established historical cause.
- **R — new reconstruction:** arithmetic or coding carried out for this packet, with method and limits.

Source identifiers link to the ledger in Section 6. Historical vocabulary is retained when identifying the object of study, not endorsed as a description of actual human value.

# 1. Verified timeline

| Date | What is supported | Use and limitation | Sources |
|---|---|---|---|
| June 18, 1873 | Vickery passed the Pharmaceutical Society’s Minor examination. | **H:** the professional museum identifies her as the first woman qualifying through this route as Chemist and Druggist. The same account records earlier female registration and a different examination route. Do not call her simply the first woman pharmacist. Original register not inspected. | S22 |
| 1877 | Vickery gave testimony in the Bradlaugh–Besant birth-control publication trial. | **H:** later biographical accounts support this; the original trial transcript has not been independently checked. Keep it attributed, or omit exact testimony until the transcript is obtained. | S23 |
| 1886 | Galton published *Regression towards mediocrity in hereditary stature*. | **F:** quantitative work on stature predates the 1904 program. This is not evidence that social convergence on an exemplar is the same statistical phenomenon. | S16 |
| 1892 | Galton’s *Finger Prints* and Pearson’s *The Grammar of Science* were published. | **F:** two concrete prior contributions, established by catalogues of original editions. Not proof that their later interpretations inherit the authority of those works. | S11, S17 |
| 1892–1893 | Weldon investigated correlated measurements in shrimp and crabs. | **F, archival:** Royal Society manuscript descriptions identify the studies. He enters the discussion with actual experience measuring biological variation. Full original manuscripts were not transcribed here. | S14–S15 |
| April–December 1897 | *The War of the Worlds* was serialized in *Pearson’s Magazine*. | **F, first-edition bibliography:** distinguish the serialization from the revised book. The magazine’s publisher was C. Arthur Pearson, not the statistician Karl Pearson. | S09–S10 |
| 1897 | Hutchison and Rainy published *Clinical Methods*. Thomson published *Cathode Rays*. | **F:** concrete clinical and physical-science context. Neither establishes what participants said about these works on May 16. | S20, S24 |
| 1898 | Heinemann published the first British book edition of *The War of the Worlds*. | **F:** Morgan copy PML 147397. Do not mix serial, first-edition, later digital, radio, film, or musical wording. | S09 |
| 1900 | Pearson published the criterion paper associated with chi-square goodness-of-fit testing. | **F, publisher record:** a dated mathematical contribution. Do not insert its formula into Galton’s construction as if it were used there. | S12 |
| 1901 | The first volume of *Biometrika* began; Hobhouse’s *Mind in Evolution* appeared. | **F / bibliographic:** quantitative biology and comparative cognition are legitimate elements of the intellectual setting. Do not call Hobhouse an LSE professor in 1904 on the strength of his later appointment. | S13, S21 |
| October 29, 1901 | Galton delivered the second Huxley Lecture to the Anthropological Institute. | **F:** origin of the numerical construction. | S04, first-page footnote |
| October 31, 1901 | The lecture appeared in *Nature* 64, 659–665. | **F:** Tables I–II are on 660; Table III is on 661. It is not a table produced during the May 1904 meeting. | S04 |
| 1903 | Shaw published *Man and Superman*. | **F:** first-edition catalogue; preface is signed Woking, 1903. Publication is not proof the work had already been staged by May 1904. | S18–S19 |
| November 8, 1903 | Wells presented a paper to the Oxford Philosophical Society. | **F:** the 1905 appendix identifies this origin and explicitly says its text was revised. | S06 |
| December 17, 1903 | Orville Wright recorded and telegraphed the powered flights. | **F:** a primary-document example of contemporary technological possibility. Do not invent the 1904 participants’ responses to the event. | S25 |
| May 16, 1904 | Galton’s paper was read to the Sociological Society at the School of Economics and Political Science, London University; Pearson chaired. | **F:** primary publication and original-offprint catalogue agree. No room layout, complete audience list, or stenographic transcript is established. | S01–S03 |
| July 1904 | The paper and associated contributions appeared in *American Journal of Sociology* 10(1), 1–25. Wells’s philosophical paper appeared in *Mind*, new series 13, no. 51, 379–393. | **F:** AJS issue date is explicit. For *Mind*, the appendix identifies no.51; modern publisher metadata inconsistently displays issue 1/January. Do not substitute that automated date for the original issue structure. | S01, S06–S07 |
| 1905 | *Sociological Papers*, vol. I, appeared, separating Discussion from Written Communications. *A Modern Utopia* reproduced a revised Wells appendix. | **F:** publication is a textual witness to the 1904 event, not the event itself. | S02, S06, S27 |

**Atmospheric context that is safe:** cite the actual works, period journal, clinical book, and flight record rather than claim that all of society shared one confidence about progress. Electricity, empire, and industrialization are broad context; any specific statistic or causal account introduced later requires its own source. Events after May 1904 must not enter participants’ minds retrospectively.

# 2. Cast, participation, and the proposition at the center

## 2.1 The publication roster—not an invented seating plan

The 1905 contents distinguish these groups. The complete roster is supplied in `data/contributors.csv`. [S02]

**Paper and reply:** Francis Galton.

**Discussion, pp52–63:** Karl Pearson; Dr. Maudsley; Dr. Mercier; Francis Warner; W. F. R. Weldon; Robert Hutchison; H. G. Wells; Mrs. Dr. Drysdale Vickery; Benjamin Kidd; Mr. Elderton; L. T. Hobhouse.

**Written Communications, pp64–78:** W. Bateson; C. S. Loch; W. Leslie Mackenzie; G. A. Archdall Reid; J. M. Robertson; George Bernard Shaw; Lady Welby.

That is **19 named contributors including Galton**, not 19 verified people physically present. Galton’s presentation and Pearson’s chairmanship have independent catalogue support. The Discussion section records oral participation; the seven written contributions have **unknown physical-attendance status**. Victor Branford’s editorial work is separately documented; he should not be silently added to the spoken roster. The book’s other authors, such as Geddes and Durkheim, are not thereby participants in this discussion. [S02–S03, S26]

The AJS transcription presents contributions in a different order and does not preserve the same clear oral/written split. Do not infer speaking sequence, simultaneous exchanges, or who stood up after whom from its headings. References to another contributor’s argument do not independently establish that contributor’s physical presence.

## 2.2 Focused cast table

The requested six plus **Hutchison and Hobhouse** provide distinct clinical and epistemic positions. Robertson remains indispensable to analysis of the key sentence and is treated in 2.4; Mackenzie and Bateson remain in the full dataset rather than receiving compulsory essay biographies.

| Person / role | Contribution already established by 1904 | What the record challenges or invokes | Uncertainty and proposed action | Avoid |
|---|---|---|---|---|
| **Francis Galton — author, presenter, reply** | Quantitative heredity research; 1886 regression paper; *Finger Prints* (1892). | Animal-breeding analogy; normal distributions; family histories; historical demography; social effects on marriage. Defines desirable qualities and differential contribution to the next generation. | Names uncertainties in heredity and the distant future; urges study and cautions against overzealous immediate action, while advocating eventual public commitment and social influence on reproduction. | Do not make his program either morally neutral data collection or an already enacted compulsory state system. S01, S04, S16–S17. |
| **Karl Pearson — chair and opening discussant** | *The Grammar of Science* (1892); 1900 criterion paper; quantitative-biological research context. | Defends founder-led scientific authority, praises Galton, and frames continuity of national mental and physical qualities as a problem. No new equation is offered in this intervention. | Says he is not a Sociological Society member and doubts what a society can accomplish without a founder; nonetheless hopes it succeeds. | Do not call him founder/member of this Society from chairmanship, or assign him authorship of Galton’s table. S01–S03, S11–S13, S28. |
| **H. G. Wells — recorded discussant** | *The War of the Worlds* serialized 1897/book 1898; scientific-romance writing; 1903 philosophical address. | Challenges professional eminence as a hereditary proxy, emphasizes individuality and social advantage, rejects reliable conscious selection of “successes.” | Nevertheless endorses sterilization of “failures.” His criticism does not amount to rejection of reproductive selection. | Do not turn him into either a uniformity advocate or an uncomplicated opponent of eugenics. S01, S06, S08–S10. |
| **George Bernard Shaw — written contributor** | Playwright and author of *Man and Superman* (1903). | Endorses eugenic improvement in civilizational language; questions marriage’s suitability for reproductive selection; contrasts actual killing with deliberate creation of future people. | Proposes childbearing under defined public conditions beyond ordinary marriage constraints, with results in a distant future and ordinary marriage continuing alongside it. | No invented speech or presence. Do not automatically equate his particular proposal with compulsory mating; do not describe it merely as a vague ethical sentiment. S01–S02, S18–S19, S30. |
| **Alice Drysdale Vickery — recorded discussant** | Physician and reproductive/women’s-rights advocate; professional museum documents her 1873 Minor examination. The 1877 trial role is supported by later history, not a trial transcript checked here. | Supports Hutchison’s attention to bad conditions; questions male-line records and asks about wives’ names when family names disappear; stresses female contribution to future generations. | Proposes improved treatment and education of women. Still uses period population-improvement language; demographic assertions in the speech are not verified statistics here. | Do not call her the first woman pharmacist without specifying qualification route, or an uncomplicated modern anti-eugenics voice. S01–S02, S22–S23. |
| **W. F. R. Weldon — recorded discussant** | Empirical quantitative studies of variation/correlation in shrimp and crabs in the 1890s. | Invokes *The Grammar of Science* to distinguish description from explanation; defends statistics; argues experimental simplification needs broader validation. Responds to Bateson’s critique of aggregate methods. | Laboratory regularity must be tested against more extensive observations before being made a maxim for human breeding. | Do not convert his claim about statistics into a timeless proof that statistical methods cannot support causal inquiry. S01–S02, S14–S15. |
| **Robert Hutchison — recorded discussant** | Coauthor with Harry Rainy of *Clinical Methods* (1897); clinical perspective. | Argues that food and environment may offer more than changing hereditary “stock”; invokes feeding and development. | Expresses confidence in nutrition, not a quantified intervention result. | No invented trial, effect size, or proof that heredity is irrelevant. Use Dr., not a later honorific copied from a catalogue. S01–S02, S20. |
| **L. T. Hobhouse — recorded discussant** | Author of *Mind in Evolution* (1901), a work on mind/evolution/comparative psychology. | Acknowledges limited biological expertise and conflicting accounts; argues the knowledge is inadequate for regulating reproduction. Environment seems more safely alterable. | Defers intervention on stock but explicitly allows that better knowledge could affect ethical judgment and legislation. | Do not turn present caution into a categorical rejection of all future intervention; no 1904 professorship claim. S01–S02, S21. |

## 2.3 Exact “best specimens” passage and its context

**F — text, not our paraphrase:**

> “The aim of eugenics is to represent each class or sect by its best specimens; that done, to leave them to work out their common civilization in their own way.”

Location: Galton’s opening argument, immediately after the nighttime animal fable and the examples of Marcus Aurelius and Adam Bede; S01. The wording is supported by the published-text transcription. Publication typography and any excerpt displayed as a facsimile must still be checked against the selected page image.

The context changes the interpretation in five ways.

1. **The stated scope is inborn qualities.** The opening definition includes development, but Galton narrows this particular address to improving a population’s inborn qualities or “stock.” It is not interchangeable with every possible environmental improvement.
2. **The fable acknowledges conflicting ideals.** Predators, prey, parasites, fish and the cuckoo do not agree about an absolute morality. The setting and dialogue are Galton’s thought experiment, not an observational study of animal preferences.
3. **Galton explicitly retains different human types.** The “best specimens” sentence is not a proposal to make everyone identical.
4. **He names criteria.** The immediately following paragraph includes health, energy, ability, manliness, and courteous disposition. Specialists assess specialist aptitudes; the community can exclude representatives it rates as undesirable. It is therefore wrong to say “best” has no supplied meaning at all. The harder question is whether those meanings are measured, commensurable, hereditary, and sufficiently justified to direct reproduction.
5. **The action is reproductive and cultural.** Galton wants “useful” classes to contribute disproportionately to the next generation, proposes family records, and considers changing social attitudes to marriage. The closing program moves from academic acceptance through practical consideration to public conscience, with an explicit warning against overzeal. [S01]

**I — the project’s defensible question:** Does allowing diversity *between* kinds settle the authority used to define and favor the preferred members *within* each kind? The historical text raises that question; it does not itself demonstrate a universal social mechanism.

## 2.4 Other voices that must stay in the analytical record

**Robertson — written.** He asks what the central sentence means and suggests Galton is making today’s “best” the standard for future averages. He disputes “large and thriving” as a general ideal and notes that commercial success can reflect capital or traits he considers antisocial. He also retains restrictive proposals about certain inherited conditions. This is criticism from within a mixed position, not a ready-made contemporary ally. [S01–S02]

**Mackenzie — written.** He distinguishes economic standing from biological qualities, invokes public health and nutrition, and proposes records and empirical tests. His comments give the dataset a route from a population description to modification of the environment, rather than only selection of reproduction. [S01–S02]

**Bateson — written.** He argues controlled breeding experiments can reveal inheritance patterns that broad aggregate statistics conceal. He does not thereby reject the paper’s ambitions. Weldon’s response creates a genuine dispute over methods, units and transport from laboratory results. [S01–S02, S29]

**Kidd — discussion.** He distinguishes individual from social efficiency and warns about scientific authority. Claims he attributes to Pearson must remain *Kidd’s report* until checked against Pearson’s original text. [S01–S02]

**Welby — written.** Her attention to women, education and future generations must not be merged with Vickery’s comments simply because both concern women. They are separate propositions by separate contributors. [S01–S02]

# 3. Proposition-level coding scheme for MRI-003

## 3.1 Corpus and unit

Use the eugenics section of **Sociological Papers I (1905), pp43–79**, as the canonical corpus, with S01 as an alternate textual witness. Segment the paper, each contribution, and Galton’s reply separately. Exclude later press notices and the adjoining investigations of kinfolk from the primary corpus. They may enter as explicitly separate contextual records.

The unit is **one proposition with an attributable evidential, explanatory, normative, or action-related function**. A sentence can contain several propositions; a proposition can extend across sentences. Each retains a pointer to the complete contribution so that qualification and context remain visible.

The supplied `proposition_seed.csv` contains **25 curated examples**, not an exhaustive sample and not a dataset from which to report frequencies. Some combined examples must be split during full segmentation. There is no completed inter-coder reliability statistic.

## 3.2 Fields

| Field group | Required fields | Rule |
|---|---|---|
| Identity | `claim_id`, `contributor_id`, `contribution_id`, `source_id`, `edition`, `record_role` | One person may make several incompatible or differently qualified claims. Do not code a whole person once. |
| Text/provenance | `page_or_locator`, `exact_span`, `context_before`, `context_after`, `source_url`, `transcription_status`, `variant_note` | Preserve original wording and spelling. Keep normalized reading text separate. Do not silently fix suspicious OCR. |
| Statement type | `claim_role` | Allow observation report, definition, causal claim, alternative explanation, analogy, evidential objection, normative criterion, proposed inquiry, proposed intervention, action restraint. Multiple linked records are preferable to a single overfull label. |
| Referent | `unit_of_reference`, `historical_term` | Individual, sex, parent pair, family, occupation, class, sect, national population, species, institution, method. Preserve uncertainty about what “race” denotes in that passage. |
| Evidence | `evidence_invoked`, `evidence_source_named`, `evidence_independently_checked`, `evidence_locator` | “Invoked” is not “validated.” Values include tables, family histories, clinical experience, experiments, examples, analogy, authority, assertion, none stated. |
| Explanation | `causal_factors`, `causal_claim_strength_as_worded` | Heredity, nutrition, social advantage, institutions, education, reproduction, multiple, unresolved, not stated. Heredity and environment are not exclusive bins. |
| Standard | `desired_outcome`, `criterion`, `criterion_operationalization` | Record the quality desired; distinguish named but unmeasured, measurable proxy proposed, operational rule, or not stated. |
| Action | `proposed_action`, `target`, `decision_authority`, `action_modality`, `timing` | Research, education, environmental change, social influence on marriage, voluntary arrangement, restriction, sterilization, deferred action, other. Do not infer compulsion solely from a reproductive topic. |
| Uncertainty | `uncertainty_kind`, `uncertainty_span`, `effect_on_action` | Data quality, mechanism, criterion, transport, feasibility, future effects. Does it halt action, narrow it, defer it, coexist with it, or have no stated consequence? |
| Relations | `responds_to_claim`, `relation_type`, `explicitness` | Edges only when documented or clearly labeled analyst interpretations. Do not construct a social-contact network from conceptual similarity. |
| Review | `coder`, `reviewer`, `agreement_status`, `adjudication_note`, `interpretation_flag` | Independent checking is an actual process, not an adjective for two models producing similar answers. |

Use separate values for **not stated**, **not applicable**, **ambiguous**, and **unverified**. These are not zeros.

## 3.3 Review and counting

Freeze the codebook after a small pilot, then code the full corpus. A second reader or independently instructed reviewer checks critical source spans, participation status, and every claim supporting a headline. Record whether reviewers are human or AI-assisted. Adjudicate disagreements rather than use model agreement as proof.

Publish all codebook amendments. Keep person-level counts and proposition-level counts distinct: a verbose contributor must not silently count as many independent participants. The corpus is an edited purposive historical record, not a random sample of public opinion. Use descriptive summaries with denominators; no significance claims about “the era” from nineteen contributors.

Do not include an ideology field, paternalism score, moral ranking, or automatic villain cluster. Ideological affiliations may be documented context elsewhere but are not an explanation assigned to each claim. Do not rank political positions or participants.

## 3.4 Questions the completed dataset can answer

- Where does disagreement occur: record coverage, explanation, desirability criterion, intervention, or timing?
- When uncertainty appears, does the particular proposition connect it to restraint or to continued action?
- Which proposed actions follow from environmental, hereditary, or mixed explanations?
- Which criteria have an operational definition, and which remain specified only in ordinary evaluative language?
- Which passages challenge the eligibility rule for the proposed family dataset?
- Which disagreements remain visible when names are hidden—and which require historical context to be understood?

The data cannot establish motives, population-level causal effects, later cultural influence, or a general law that shared concern inevitably produces control. Similarities revealed after name removal do not by themselves prove that ideology is causally irrelevant.

## 3.5 The public representation

A useful minimal interface is **Claim / Evidence invoked / Criterion / Proposed action / Qualification**, with a visible source link. Begin with the full range of claims around Galton’s proposition. Removing names changes the view, not the dataset. It should not force historically different arguments into predetermined camps.

Numerical material appears as one source used in the wider argument. Do not attach a modern table-reconstruction result to a contributor unless the record actually connects them.

# 4. Galton numerical reconstruction specification

## 4.1 What the source is

S04 is **“The Possible Improvement of the Human Breed under the Existing Conditions of Law and Sentiment,” Nature 64 (October 31, 1901), 659–665**, based on the lecture of October 29. It is explicitly named in the 1904 paper’s list of relevant work. This provides a documented link between the meeting and the earlier numerical object; it is not an invented thematic pairing. [S01, author’s bibliography; S04]

The construction has three distinct evidential objects:

1. **A theoretical normal distribution.**
2. **A comparison with social categories attributed to Booth, after approximation and regrouping.**
3. **A theoretical intergenerational transition model.**

Do not collapse these into “Galton measured ten thousand people’s civic worth.” No such dataset is supplied by these tables.

## 4.2 Table I: the normal classes — reproducible now

The unit is Galton’s **normal-talent**, corresponding to the normal distribution’s probable error, not one standard deviation. With a standard normal variable Z:

`q = Phi_inverse(0.75) = 0.6744897501960817`

Class boundaries are `-infinity, -4q, -3q, -2q, -q, 0, q, 2q, 3q, 4q, infinity`.

For bounds a and b, the theoretical count is `10000 * (Phi(b) - Phi(a))`.

| Class, lower to higher | Printed per 10,000 | Modern integral, rounded |
|---|---:|---:|
| v and below | 35 | 35 |
| u | 180 | 180 |
| t | 672 | 672 |
| s | 1,613 | 1,613 |
| r | 2,500 | 2,500 |
| R | 2,500 | 2,500 |
| S | 1,613 | 1,613 |
| T | 672 | 672 |
| U | 180 | 180 |
| V and above | 35 | 35 |

**R — executed check:** all ten rounded entries match; the total is 10,000. This reproduces theoretical arithmetic. It does not validate a normal distribution of civic worth.

Galton explicitly acknowledges Pearson’s work on departures from symmetry and warns against blindly trusting a normal law. He nevertheless adopts normality as an approximation for this construction. That qualification must appear alongside the curve. [S04, p659]

## 4.3 Table II: Booth comparison — preserve each transformation

S04 p660 reports the following per-thousand values. They are **Galton’s published Booth-derived quantities**, not independently recovered raw Booth observations.

| Combined group | Galton’s reported rate | His approximation |
|---|---:|---:|
| H: all above G | 97 | 100 |
| F + G | 200 | 200 |
| E | 382 | 400 |
| C + D | 227 | 200 |
| A + B | 94 | 100 |
| **Total** | **1,000** | **1,000** |

The source’s social categories combine income/occupational descriptions with evaluative language. That mixture is part of the object of analysis; it is not a modern measurement scale.

Galton explains that Booth’s all-London groupings are not identical to the pairings he wants. He disentangles them using proportions from East London and says the result is approximate. He also observes that class boundaries blend. [S04, pp660–661]

The displayed regrouping is:

- H 100 → upper tail 100.
- F+G 200 → S 150 + R 50.
- E 400 → R 200 + r 200.
- C+D 200 → r 50 + s 150.
- A+B 100 → lower tail 100.

This produces `[100,150,250,250,150,100]`, compared with the displayed normal proportions `[89,161,250,250,161,89]`.

**R — executed check:** the differences are `[11,-11,0,0,-11,11]` per thousand. The absolute differences sum to 44; their total-variation distance is 0.022. This is only a compact description of agreement **after the published transformations**. It is not a goodness-of-fit test, evidence for hereditary civic value, or proof of deliberate manipulation.

**Do not use 1,000 as a sample size.** These are normalized rates. A chi-square test would require an actual sampling basis, denominators, dependence structure, and an account of the fitted choices.

### Upstream gate

Locate the exact Booth edition/table and reproduce the East-London-to-all-London disaggregation before labeling the initial quantities “independently verified Booth data.” The source route is the LSE Booth collection and pre-1901 published survey volumes. That tracing is unresolved in this packet. No invented source page or imputed individual records may fill the gap. [S31]

### Defensible sensitivity checks

Keep Galton’s original construction as the reference. Separate sensitivity to rounding from sensitivity to group definitions. An identity view that leaves original reported rates untouched is informative but is not automatically an alternative mapping to the same six normal bins. Any altered crosswalk must have a stated reason. Record what is hypothetical. Do not tune cutpoints until agreement is maximized, then present the resulting fit as independently discovered.

## 4.4 Table III: descent — what is assumed

The table is headed with three conditions: parents alike in class, population statistics continuing unchanged, and normal frequency law applicable throughout. The prose adds equal fertility, a simplified equal-sex outcome, no account of death before marriage, and one adult son and daughter per parental pair in the idealized population. Male and female classifications are treated as corresponding scales. [S04, p661]

These are **stipulations**, not conclusions inferred from observed parent–child civic data.

The verbal procedure, corroborated in the 1909 reprint, concentrates each parental class at its within-class median, moves the filial center one-third toward the population center, and disperses offspring with approximately three-quarters of the parental spread. [S04; S05, pp15–18]

A modern literal specification, clearly labeled as such, is:

`m_i = Phi_inverse((Phi(a_i) + Phi(b_i))/2)`

`mu_child_i = (2/3) * m_i`

`expected_ij = N_i * [Phi((b_j - mu_child_i)/0.75) - Phi((a_j - mu_child_i)/0.75)]`

The tail medians are finite because each tail is represented by its probability median. Do not replace them with an arbitrary finite class midpoint.

### What has actually been checked

- The packet preserves all 100 displayed cells, including blanks separately from their treatment as zero in arithmetic sums.
- Summing the printed cells reproduces printed descendant margins of **34, 168, 655, 1,623, 2,522, 2,522, 1,623, 655, 168, 34**, totaling **10,004**.
- The printed row-label parental counts total 10,000. The source notes approximate agreement and discrepancy associated with coarse classes. A small discrepancy is not a warrant to call the table fabricated.
- Some printed parental class counts differ from Table I, and the upper header’s T entry appears as 672 while the corresponding row label is 671. Preserve the facsimile reading and require independent review before highlighting the typographic discrepancy publicly.
- The literal modern specification does **not** reproduce every historical cell. In the supplied implementation, 42 rounded cells differ, by at most 10 per cell. Residuals are supplied. Do not relabel this as exact reproduction or optimize undocumented parameters to eliminate them.

A useful mathematical check: for a continuous linear-normal stationary model standardized to variance 1, `beta^2 + s^2 = 1`. The convenient fractions `beta=2/3` and `s=3/4` give `145/144`, not exactly 1. The corresponding exact spread would be `sqrt(5)/3 ≈ 0.745356`. This is a check on a modern interpretation of the verbal model, **not an assertion that it identifies the sole reason for every discrepancy in Galton’s discrete approximation**. Do not quietly replace the historical 0.75.

### Claims derived from the construction

The table’s high-class yield comparisons lead to claims about the relative economy of encouraging selected parentages. One example uses `(6/35)/(3/2500) = 142.857...`, approximately 143. This is a ratio **inside the stipulated model**, not an observed finding that one kind of person is worth 143 of another. [S04; S05, “Economy of Effort”]

Table IV compares one versus two class-V parents, changing the regression and spread. It is a later reproduction target, not part of the completed scan-level checks. Do not build its final interactive until the original page and all its entries are verified.

## 4.5 Analytical outputs and display

The code in `code/check_galton.py` uses only Python’s standard library. It reads the preserved CSVs and writes reproducible tables and `results/arithmetic_checks.json`.

Use four distinct display labels: **originally printed / newly reproduced / hypothetical alternative / not established by these data**. Judgments entered at the initial construct and grouping stages, not only after the arithmetic. Avoid a false “pure evidence above / judgment below” divider.

A suitable final question is: **What does this calculation establish under these assumptions—and what further evidence would its proposed application require?**

# 5. Wells visual-representation specification

## 5.1 Source and precise argument

Use S06, the revised appendix to *A Modern Utopia*. Its origin note identifies November 8, 1903, Oxford Philosophical Society, and a revised *Mind* publication. Cite the version actually quoted, not a supposed verbatim transcript of the 1903 talk. [S06–S07]

The key passage begins **“Let me give you a rough figure”** and describes black-and-white reproduction using a rectangular net. From farther away the image seems faithful; close viewing reveals uniformly shaped rectangles rather than the original forms. Wells then considers finer classification and the limits of the instrument.

Retain his qualification: categories can work exceedingly well for practical purposes, and inquiry can check and correct its adjustments. His argument is not simply “all classification is useless.” Later in the appendix he also explains how his skepticism coexists with firm commitments. Do not portray the disagreement between provisional knowledge and action as something he never considered. [S06]

## 5.2 What to build

**One modern demonstration, not an invented recovered artifact.** Choose a rights-cleared reference image with meaningful detail. Produce a regular rectangular sampling grid. Show three views: reference image; fixed coarse-grid reconstruction; finer-grid reconstruction.

Two controls have different meanings:

- **Viewing distance / magnification** reveals a fixed representation’s existing structure.
- **Resolution** changes the amount of information preserved in the representation.

Do not use zoom as a deceptive animation that silently changes the sampled data. Keep the reference crop, tonal mapping and comparison scale explicit.

A small reveal can show a distinction preserved at one resolution and lost at another, but it must be a property of the chosen demonstration—not a fabricated finding about actual historical populations. No universal “truth percentage” or claim to empirically prove philosophical uniqueness.

## 5.3 Appearance

Contemporary white or pale-neutral canvas; crisp linework; ample space; restrained modern type. The period influence is the **diagram’s logic**, not antique decoration. No parchment, sepia, gears, brass, fake typewriter lettering, or ornamental frame. Rectangles are central because Wells specifies them. Circles appear only if required by a separate analytical relationship.

The historical original can be opened as a source, not used as a textured wallpaper. Interaction must work on mobile and have static panels, keyboard access and explanatory alternative text.

## 5.4 How it relates to the claims and numbers

The Wells display makes representation visible. It does not prove Galton’s data false. The Galton audit tests a particular construction. It does not prove all classification fails. Keep these evidential jobs separate while sharing a visual grammar.

For the 1904 exchange, a typographic field of propositions or a simple claim matrix is preferable to an invented seating chart. Written contributors must remain visibly distinguished from reported speakers, including when names are hidden.

## 5.5 The small literary source bank

| Passage | What the text supplies | What the essay may interpret | What it does not establish |
|---|---|---|---|
| *War of the Worlds*, Book I.1 | Humans watched as humans watch microscopic organisms; assumed superiority; a narrator comparing Martian treatment of humans with human treatment of others. | Apparent standing depends on the perspective and conditions of observation. | A literal prediction of AI, current biology, or historical accuracy of every assertion in the narrator’s voice. |
| Book II.7, “The Man on Putney Hill” | The narrator briefly defers to the artilleryman’s situational judgment; the latter proposes a selective underground future; the narrator becomes disillusioned with the gap between scheme and completed work. | A reversal of an old hierarchy does not establish the new claimant’s competence or entitlement. | The artilleryman’s program is not automatically Wells’s position. |
| Book II.8, “Dead London” | The narrator attributes the Martians’ death to terrestrial putrefactive and disease bacteria; technological superiority is insufficient. | A restrained literary echo of fitness under conditions. | This is not contemporary immunology, a COVID comparison, or empirical proof of a general thesis. |
| Book II.10, epilogue | The narrator qualifies what is known about Martian biology. | Even the explanatory ending retains uncertainty. | Do not erase this qualification by treating all fictional scientific explanations as omniscient authorial fact. |

All four entries derive from S08. Use at most the passages genuinely needed. No Orson Welles radio-panic excursus, contemporary AI cast, or future-war causal chain belongs in this packet. Richard Burton/Jeff Wayne is a separate, later auditory layer; no recording or music is cleared for reuse by a nineteenth-century literary source.

# 6. Essay source ledger and verification priorities

The complete machine-readable ledger is `sources/source_ledger.csv`. Each entry specifies its type and access limitation. Bibliographic catalogue verification is not represented as full-page transcription; modern institutional biography is not relabeled primary testimony.


## 6.1 Source-by-source ledger

**S01 — [Galton, Eugenics: Its Definition, Scope, and Aims; discussion and reply](https://galton.org/essays/1900-1911/galton-1904-am-journ-soc-eugenics-scope-aims.htm)**  
1904 · American Journal of Sociology 10(1), 1–25; DOI 10.1086/211280  
Type: Primary text in later web transcription. **Check status:** Text examined. Heading order differs from the 1905 proceedings; transcription has errors. Use for arguments, not an attendance list.

**S02 — [Sociological Papers, volume I: contents and eugenics section](https://wellcomecollection.org/works/ds624dzu/items?canvas=11&page=2&shouldScrollToCanvas=true)**  
1905 · Paper 43–51; Discussion 52–63; Written Communications 64–78; reply begins 78  
Type: Digitized primary volume; catalogue and contents. **Check status:** Contents examined; distinguishes spoken discussion and written communications. No independent audience register inspected. Site currently labels the item In copyright; do not assume its scans are cleared for reuse.

**S03 — [Galton offprint catalogue: Eugenics: Its Definition, Scope, and Aims](https://wellcomecollection.org/works/vf2e8j9k)**  
1904 · GALTON/2/13/3/10; original at UCL Special Collections  
Type: Archival catalogue of primary document. **Check status:** Confirms May 16, 1904, School of Economics and Political Science, London University, Pearson in chair.

**S04 — [Galton, Possible Improvement of the Human Breed under the Existing Conditions of Law and Sentiment](https://pdf.galton.org/essays/1900-1911/galton-1901-improvement.pdf)**  
1901 · Nature 64, 659–665, October 31; lecture October 29  
Type: Primary facsimile. **Check status:** PDF text examined; page images 660 and 661 inspected. Tables I–III transcribed in this packet. Full descent-figure and Table IV image verification incomplete.

**S05 — [Galton, Essays in Eugenics: 1901 lecture reprinted](https://www.gutenberg.org/files/73802/73802-h/73802-h.htm)**  
1909 · Chapter I; printed-image links and text, especially pp15–19  
Type: Later authorial reprint in digital transcription. **Check status:** Corroborates regression, medians and spread description. Do not silently treat reprint typography as identical to 1901.

**S06 — [Wells, Scepticism of the Instrument](https://www.marxists.org/reference/archive/hgwells/1905/modern-utopia/appendix.htm)**  
1905 · A Modern Utopia, Appendix; origin note identifies Oxford talk November 8, 1903 and revised Mind version  
Type: Primary text in digital transcription. **Check status:** Full appendix examined. Rectangular image analogy and practical-utility qualifications verified. No actual original image accompanies the passage.

**S07 — [Wells, Scepticism of the Instrument, original journal record](https://academic.oup.com/mind/article-abstract/XIII/1/379/1292046)**  
1904 · Mind, new series 13, no.51, 379–393; DOI 10.1093/mind/XIII.1.379  
Type: Original publisher bibliographic record. **Check status:** Article record inspected, full PDF not retrieved. Website labels issue/date as issue 1/January; Wells’s own appendix specifies no.51. Cite year and original issue number, not website January metadata.

**S08 — [Wells, The War of the Worlds](https://www.gutenberg.org/files/36/36-h/36-h.htm)**  
1898 work; digital edition · Book I.1; Book II.7–8; Book II.10  
Type: Primary literary text in digital edition. **Check status:** Relevant passages examined. Do not assume this transcription is textually identical to the 1897 serial or every 1898 printing.

**S09 — [The War of the Worlds, first edition catalogue and copy](https://www.themorgan.org/printed-books/371751)**  
1898 · Morgan Library PML 147397; London: William Heinemann  
Type: Catalogue of primary first edition. **Check status:** Verifies 1898 edition and April–December 1897 Pearson’s Magazine serialization.

**S10 — [Wells manuscript and revisions for book publication](https://www.library.illinois.edu/rbx/hgwells2016/2016/09/87/)**  
circa 1897 · University of Illinois Rare Book & Manuscript Library, WOW-001  
Type: Custodian description of primary manuscript. **Check status:** Identifies manuscript and revisions; useful source for edition history, not proof of claims made by fictional narrator.

**S11 — [Pearson, The Grammar of Science](https://catalogue.nli.ie/Record/vtls000454935)**  
1892 · London: Walter Scott; New York: Charles Scribner’s Sons  
Type: National Library of Ireland catalogue of primary book. **Check status:** Bibliography verified. A full textual edition audit has not been performed here.

**S12 — [Pearson, On the criterion that a given system of deviations ...](https://doi.org/10.1080/14786440009463897)**  
1900 · Philosophical Magazine, series 5, 50(302), 157–175; DOI 10.1080/14786440009463897  
Type: Original publisher bibliographic record. **Check status:** Article identified; direct publisher page intermittently inaccessible. Supports dated contribution, not reconstruction of this paper.

**S13 — [The Scope of Biometrika](https://doi.org/10.1093/biomet/1.1.1)**  
1901 · Biometrika 1(1), 1–2; DOI 10.1093/biomet/1.1.1  
Type: Original journal record. **Check status:** Opening journal article identified. Do not assign every editorial statement to one person without checking signatures.

**S14 — [Weldon, Certain correlated variations in Crangon vulgaris](https://makingscience.royalsociety.org/items/pp_19_1/paper-certain-correlated-variations-in-crangon-vulgaris-common-shrimp-by-walter-frank-raphael-weldon)**  
1892 · Royal Society manuscript PP/19/1; received February 11, read March 3  
Type: Royal Society catalogue/abstract of primary manuscript. **Check status:** Indexed archival description examined; full page intermittently inaccessible. Identifies quantitative study of correlated shrimp measurements.

**S15 — [Weldon, On certain correlated variations in Carcinus maenas](https://makingscience.royalsociety.org/items/pp_22_22/paper-on-certain-correlated-variations-in-carcinus-maenas-common-crab-by-walter-frank-raphael-weldon)**  
1893 · Royal Society manuscript PP/22/22  
Type: Royal Society catalogue/abstract of primary manuscript. **Check status:** Indexed record describes Naples/Plymouth samples of adult female crabs; manuscript pages not inspected.

**S16 — [Galton, Regression towards mediocrity in hereditary stature](https://galton.org/essays/1880-1889/galton-1886-jaigi-regression-stature.pdf)**  
1886 · Journal of the Anthropological Institute 15, 246–263; lecture work reported in 1885  
Type: Primary facsimile. **Check status:** PDF identified and text inspected. This statistical concept must remain distinct from social movement toward an exemplar.

**S17 — [Galton, Finger Prints](https://wellcomecollection.org/works/echgq5v7)**  
1892 · London: Macmillan; Wellcome/King’s College London copy  
Type: Catalogue of primary book. **Check status:** Publication verified. Do not call Galton the sole inventor of fingerprint identification.

**S18 — [Shaw, Man and Superman: A Comedy and a Philosophy](https://www.themorgan.org/printed-books/86442)**  
1903 · Morgan Library PML 136171; Westminster: Archibald Constable  
Type: Catalogue of primary book. **Check status:** 1903 publication verified; publication does not mean it had already been staged by May 1904.

**S19 — [Shaw, Man and Superman](https://www.gutenberg.org/cache/epub/3328/pg3328-images.html)**  
1903 work; digital edition · Prefatory letter signed Woking, 1903  
Type: Primary literary text in digital edition. **Check status:** Prefatory date checked. Characters’ speeches are not automatically Shaw’s personal propositions.

**S20 — [Hutchison and Rainy, Clinical Methods](https://wellcomecollection.org/works/qva8ayju)**  
1897 · London: Cassell; Royal College of Physicians of Edinburgh copy  
Type: Catalogue of primary book. **Check status:** Authorship and date verified. Modern catalogue adds later honorifics; do not use them in the 1904 scene.

**S21 — [Hobhouse, Mind in Evolution](https://obnb.uk/p12688966-mind-in-evolution)**  
1901 · London: Macmillan; 1998 facsimile catalogue  
Type: Bibliographic record of facsimile of primary work. **Check status:** 1901 original publication identified; original title page not independently inspected.

**S22 — [Women Pharmacists in the 19th Century; Alice Vickery](https://membership.rcpharm.org/about-us/museum/online-exhibitions/celebrating-women-in-pharmacy/women-pharmacists-in-the-19th-century)**  
Institutional retrospective · Royal Pharmaceutical Society museum exhibition, currently Royal College of Pharmacy site  
Type: Later institutional historical account. **Check status:** Specifies Minor examination June 18, 1873. Also identifies earlier female registrants/Modified examination. Do not generalize to first woman pharmacist. Page contains an inconsistent five-years-later phrase; use its explicit date with attribution.

**S23 — [Alice Vickery (1844–1929)](https://heritage.humanists.uk/alice-vickery/)**  
Later historical account · Humanist Heritage  
Type: Later biographical interpretation. **Check status:** Connects naming, medical education and 1877 birth-control testimony. Original trial transcript and exam register not inspected; keep these biographical claims attributed.

**S24 — [Thomson, Cathode Rays](https://doi.org/10.1080/14786449708621070)**  
1897 · Philosophical Magazine 44(269), 293–316; DOI 10.1080/14786449708621070  
Type: Original paper record. **Check status:** Original publication identified. Useful period context, not evidence any participant discussed this paper at the meeting.

**S25 — [Orville Wright diary and telegram](https://www.loc.gov/collections/wilbur-and-orville-wright-papers/articles-and-essays/collection-highlights/first-flight/)**  
1903-12-17 · Library of Congress, Wright Brothers Papers  
Type: Custodian presentation of original documents. **Check status:** Dated contemporary flight records. Telegram says 57 seconds, usually contrasted with 59-second historical account; no duration claim needed here.

**S26 — [Sociological Papers publication correspondence](https://wellcomecollection.org/works/xvmfdg39)**  
1904–1905 · GALTON/2/4/19/9/2; Victor Branford letters and proofs  
Type: Archival catalogue. **Check status:** Catalogue mentions preparations, publication and corrected proofs. Correspondence itself not read; not an independent attendance register.

**S27 — [Sociological Papers online contents](https://journals.sagepub.com/toc/sora/sp1/1)**  
1905 volume, modern metadata differs · Volume SP1; paper/discussion/written communications separately indexed  
Type: Publisher contents/metadata. **Check status:** Page divisions corroborated; online January 1904 indexing is not the publication date of the physical 1905 volume or date of the May meeting.

**S28 — [Pearson opening of Discussion](https://galton.org/cgi-bin/searchImages/galton/search/essays/pages/galton-1905-socpapers-eugenics-definition-scope-aims_5.htm)**  
1905 · Sociological Papers I, 52–53  
Type: Page-specific approximate OCR of primary text. **Check status:** Text examined; clearly marked approximate OCR. Includes editorial cross-reference to Branford’s reply.

**S29 — [Bateson written communication and following comments](https://galton.org/cgi-bin/searchImages/galton/search/essays/pages/galton-1905-socpapers-eugenics-definition-scope-aims_11.htm)**  
1905 · Sociological Papers I, 64–65  
Type: Page-specific approximate OCR of primary text. **Check status:** Heading Written Communications checked in indexed page. Quotes should be checked against page images before publication.

**S30 — [Shaw written communication](https://galton.org/cgi-bin/searchImages/galton/search/essays/pages/galton-1905-socpapers-eugenics-definition-scope-aims_16.htm)**  
1905 · Sociological Papers I, 74–75  
Type: Page-specific approximate OCR of primary text. **Check status:** Indexed excerpt examined; direct page intermittently inaccessible. Use S02 for written status and S01 for content.

**S31 — [Charles Booth archive](https://www.lse.ac.uk/library/collection-highlights/charles-booth)**  
1880s–1900s · London School of Economics Library collection  
Type: Archival collection finding aid. **Check status:** Upstream source route only. Exact Booth edition/page and Galton’s East-London-to-all-London calculation remain unverified.


## 6.2 Preferred citation practice

For the meeting’s setting and recorded participation, cite the original offprint description and the 1905 volume’s contents; for a contributor’s argument, cite the passage in the chosen edition. Do not cite the contents alone as proof of the content of a speech. For the numerical work, cite the **1901 original page** and identify the new reproduction separately. For Wells, distinguish the 1905 revised appendix from the date of the preceding oral paper.

Stable bibliographic citations:

- Francis Galton, “Eugenics: Its Definition, Scope, and Aims,” *American Journal of Sociology* **10**, no. 1 (July 1904): 1–25, including discussion and reply. DOI: [10.1086/211280](https://doi.org/10.1086/211280).
- “Eugenics: Its Definition, Scope and Aims,” in *Sociological Papers*, vol. I (London: Macmillan, 1905), paper 43–51, Discussion 52–63, Written Communications 64–78, reply beginning 78. The physical volume, not inconsistent current index metadata, establishes its publication date.
- Francis Galton, “The Possible Improvement of the Human Breed under the Existing Conditions of Law and Sentiment,” *Nature* **64** (October 31, 1901): 659–665. The first-page footnote dates the Huxley Lecture to October 29.
- H. G. Wells, “Scepticism of the Instrument,” *Mind*, new series **13**, no. 51 (1904): 379–393; revised as the appendix to *A Modern Utopia* (1905). DOI: [10.1093/mind/XIII.1.379](https://doi.org/10.1093/mind/XIII.1.379). Preserve the issue/date metadata warning in S07.
- H. G. Wells, *The War of the Worlds* (London: William Heinemann, 1898). The Morgan first-edition record distinguishes it from the 1897 serialization. The linked Gutenberg text is a reading witness, not proof that all its wording matches the first issue of the serial.

## 6.3 Remaining source gates

1. Obtain the exact upstream Booth tables and reproduce the disaggregation behind Galton’s reported quantities. Catalogue access to the Booth collection is not completion of this task.
2. Obtain a stable full facsimile of the 1905 eugenics section and align every seed proposition to the canonical printed page. The published web transcription is useful but contains textual errors.
3. Independently review the numerical transcription, especially Table III’s 671/672 and 1613/1614 differences and the printed blanks.
4. Resolve the precise historical integration/rounding convention behind Table III before claiming an exact reproduction. The current standard-normal implementation remains a documented approximation with residuals.
5. Verify the complete descent figure and Table IV against original page images before adding them to a public interactive.
6. Obtain Vickery’s original examination/register and trial record if making detailed priority or testimony claims, and resolve “Mr. Elderton” against an authority record before expanding the name.

These are concrete limitations on particular claims. They do not invalidate the verified participation distinction, the reproduced normal-probability table, the visible Table II transformations, or the source-grounded interpretive questions.

# 7. Factual veto list

The following formulations should not enter the essay, MRI page, captions, or agent prompts without the specified correction.

| Do not write or imply | Use instead / reason |
|---|---|
| “Shaw stood up and addressed the room.” | His contribution is printed as a **Written Communication**. Attendance is not independently established. |
| “All nineteen named contributors were in the room.” | Nineteen names belong to the published paper/discussion/written-response record. It is not an audience register. |
| “Vickery is absent from the discussion.” | She is listed under the form “Mrs. Dr. Drysdale Vickery.” |
| “Vickery was the first woman pharmacist.” | Specify the Minor-examination/Chemist-and-Druggist distinction and attribute the professional museum. Earlier women were registered through other routes. |
| “Pearson was a member or founder of this Sociological Society.” | He explicitly says he is not a member; chairmanship is separately established. |
| “The 1904 discussion created Galton’s Booth table.” | The numerical lecture is from October 1901; Galton cites it in the later paper. |
| “Galton measured the civic worth of ten thousand people.” | Table I is a theoretical normal distribution expressed per 10,000. Table II’s per-thousand values are normalized comparisons, not a sample of exactly 1,000 people. |
| “Booth’s raw data have been independently reproduced here.” | Only Galton’s published Booth-derived quantities and transformations have been transcribed. The original upstream comparison is unresolved. |
| “Galton concealed every approximation.” | The table and prose make approximations and regrouping explicit. The question is what they support. |
| “Best has no stated meaning.” | Galton names desirable qualities and institutional selectors. What is not established is a common validated measurement scale and authority for the intended reproductive application. |
| “Galton wanted everyone to be identical.” | He explicitly allows diverse types and ideals. Analyze selection within types rather than erase the pluralism. |
| “Wells proves that the 1901 lecture contained no plural criteria.” | That is Wells’s 1904 characterization of the change. The 1901 ‘Diplomas’ section itself refers to alternative definitions and many forms of civic worth; distinguish this from Wells’s more specific claim about separate breeding systems by type. |
| “Wells opposed eugenics because he understood individuality.” | He challenges positive selection of successes while advocating sterilization of failures. Preserve both propositions. |
| “Hobhouse ruled out such intervention on principle for all time.” | He conditions his restraint on inadequate knowledge and allows that improved evidence could change later ethical or legislative judgment. |
| “Environmental explanations were absent.” | Hutchison, Vickery, Mackenzie and others supply counterevidence. Their inclusion does not itself prove their empirical claims. |
| “The skeptics all rejected intervention, while the confident speakers all favored it.” | This is a coding question, not an established result. No comparative counts are reported from the curated seed file. |
| “The elegant mathematics proves the social meaning.” | Reproducibility establishes the calculation under specified assumptions. The construct and application require further justification. |
| “The small Table III discrepancies prove deception.” | Galton acknowledges approximation. The present reproduction is incomplete; source typography, integration and rounding remain to be checked. |
| “A V-class person is worth 143 other people.” | Approximately 143 is a conditional modeled offspring-yield ratio in a particular comparison, not observed human value. |
| “Statistical regression is the same mechanism as convergence on a social exemplar.” | They are different claims. The relationship must be argued and tested, not supplied by a shared word. |
| “All judgment enters after the mathematics.” | Construct choice, population, grouping, approximation and model selection already involve judgments. Do not design a misleading pure-evidence/impure-judgment divider. |
| “A claim-map of this record proves a universal human tendency.” | This is one edited historical corpus. Counts describe it, not all of society or human nature. |
| “The 1904 meeting led directly to the twentieth-century wars or regimes.” | No such causal history is established by these documents. Do not use later events as evidence of participants’ knowledge or a single common lineage. |
| “Their ideological identities explain their positions.” | Reconstruct what each claim asserts, invokes and proposes. Preserve cross-cutting differences without moral or political rankings. |
| “Pearson’s Magazine means Karl Pearson’s magazine.” | The magazine’s Pearson is C. Arthur Pearson. Do not manufacture another connection from the surname. |
| “Wells left a diagram demonstrating that no representation can be adequate.” | He supplies a verbal image analogy and recognizes practical usefulness. A modern visualization is our demonstration, not his recovered artifact. |
| “The artilleryman’s scheme is Wells’s own final program.” | It is a fictional character’s proposal and the narrator becomes disillusioned with it. |
| “The Martian explanation proves a modern biological or social claim.” | Use it as fiction and interpretation. The epilogue itself qualifies what is known. Avoid repeating obsolete racial-extinction assertions in the narrator’s voice as historical fact. |
| “Original literary publication clears later music and imagery.” | The musical recording is a separate protected work. Historical text and scan-reuse status must also be distinguished. |
| “The new dataset and all calculations are independently verified.” | The roster and sources have been checked; the seed propositions and numerical transcription have had one research pass. Full independent adjudication is still pending. |

# 8. Five findings with genuine narrative potential

Each finding has a documentary basis and a separate possible reading. None is an essay draft.

## Finding 1 — Galton’s proposal includes pluralism, not just uniformity

**Documentary basis:** The animal fable begins with conflicting ideals. The central sentence retains different classes and sects; the following paragraphs name qualities and selectors. Robertson’s written response directly asks whether today’s “best” is being made the standard for future averages. [S01; S02]

**What makes this surprising:** A more differentiated representation is already part of the proposal. The problem cannot be explained simply by claiming that Galton forgot diversity.

**Interpretive opportunity:** Ask whether allowing different types settles who can define their preferred representatives and what that selection authorizes. This is a question about the criterion and its use—not a finding that all classification is coercive.

## Finding 2 — The proposed dataset already selects a particular image of success

**Documentary basis:** Galton’s proposed record of “large and thriving” families requires at least **three adult male children** and defines thriving partly through children’s superior positions compared with early classmates. He also imagines a “golden book” honoring such families, and says collecting the records will help make eugenics publicly recognizable as serious science. [S01, research program, item 3]

**What makes this surprising:** The family database is not a neutral census of all possible outcomes. Eligibility, success, recognition and institutional legitimacy are already connected in the proposed collection procedure. Vickery’s questions about women and family names make omissions in genealogical representation a contemporary issue, not just a modern criticism. [S01, Vickery]

**Interpretive opportunity:** Show the proposed form before showing any calculations. What kind of family can enter it? What would remain invisible? Do not imply the golden book was completed, or that the suggested feedback has been measured.

## Finding 3 — Criticism of the evidence and restraint in action do not map neatly onto people

**Documentary basis:** Wells challenges eminence as a hereditary proxy and rejects reliable conscious selection of successes, but proposes sterilizing failures. Hobhouse says the current state of knowledge does not justify regulating reproduction, while allowing that improved knowledge could change future judgment. Hutchison favors improving the environment; other contributions retain different forms of intervention. [S01; S02]

**What makes this surprising:** “Supports eugenics” and “criticizes the evidence” are inadequate single labels for the recorded positions. The same contribution can contain skepticism about one inference and confidence about another.

**Interpretive opportunity:** Code propositions and the effect of a qualification on the proposed action. Do not decide beforehand that uncertainty is always ignored or always effective.

## Finding 4 — The normal arithmetic works; the observational correspondence is constructed

**Documentary basis and new reconstruction:** Table I’s ten normal probabilities reproduce after rounding. Table II reaches its comparison through printed approximations and regroupings. The fitted-looking result therefore has a different evidential status from an independently sampled normality test. Table III then moves into a stipulated descent model. [S04–S05; executed results]

**What makes this surprising:** The public artifact can show working mathematics rather than seek an arithmetic scandal. Agreement at one stage does not establish every interpretation that follows.

**Interpretive opportunity:** Let the reader see the theoretical curve, then inspect the crosswalk used to make a social classification comparable to it. The further gap between a fit and a program of improvement is a separate question. Preserve the original model’s acknowledged approximations and present residuals honestly.

## Finding 5 — Wells had already explained the visual problem, and he also confronted the need to act

**Documentary basis:** The appendix traces its origin to November 1903, before the meeting. It describes the rectangular construction beneath a seemingly continuous picture, discusses finer classifications, recognizes practical success, and later addresses how skepticism coexists with firm commitments. [S06]

**What makes this surprising:** The thinker is not simply unaware that his instrument is provisional. The challenge is how provisional knowledge is used without either granting it unlimited authority or making action impossible.

**Interpretive opportunity:** Use the image as an instrument the reader can inspect, then return to the discussion and its proposed actions. Do not turn Wells’s visual analogy into a theorem proving the project’s general hypothesis.

# 9. What is ready to hand to the next stage

The historical narrative now has a checked date, a corrected publication cast, substantive disagreements, an exact central proposition, and dated cultural/scientific anchors. MRI-003 has a corpus boundary, codebook, seed records, transcribed tables, executable arithmetic and a visual specification.

The next production gate is **not more general brainstorming**. It is (a) exact page alignment and independent review of the historical claims; (b) the Booth-source and Table III reproduction work; and (c) a minimal modern prototype that labels the different evidence states correctly. The first publicly defensible arithmetic display can already be built around Table I and the explicitly attributed Table II transformations. It must not claim that the unresolved upstream values or complete descent model have been independently reproduced.

## Three questions Part I leaves open

1. When a representation influences subsequent opportunities, what observations would distinguish genuine predictive accuracy from a pattern partly maintained by those choices?
2. Which kinds of contrary evidence revise a governing representation, and which are lost because the record was not designed to contain them?
3. Can additional detail improve an abstraction without granting it authority beyond the decision it has actually been tested to support?

No modern company analysis, new Part II protocol, public deployment, or paid action is authorized or claimed by this packet.