What the Four Papers Say Together
The four papers are methodological readings of a single corpus, not four independent exercises in inductive corroboration. Each states this explicitly, as does the programme’s Common Appendix. What unites them is therefore not an accumulation of different kinds of evidence, but the disaggregation of a single question across four sites within the system for defining the entitlement-holder.
The unifying thesis is that the modern state neither moved the social recipient from charity to a right along a linear path nor created the entitlement-holder out of nothing; instead, it redistributed powers: (defining entitlement-holders, substantiating their entitlement, counting them, and reviewing decisions concerning them).
This thesis can be read at four sites: (1) In the record: the reference point for framing shifted from the founder’s stipulation to codified law, while the document embodying the right remained in place. (2) In the standard: the locus of rule-making moved from the individual deed to the general legal provision, while the claim concerning the source remains a hypothesis that has yet to be coded. (3) In counting: the units of counting entitlement-holders multiplied, making it impossible to combine them under a single denominator, and the capacity for exhaustive enumeration became part of the legal characterisation of the right. (4) In claims: the claims bundle already existed with all its elements before 1946, so the proposition gets the starting point wrong even before it gets the direction wrong.
And What They Do Not Say
The papers do not claim that the modern side of these transitions has been established: their material is a corpus of fatwās (juristic opinions), and its testimony about the period before the transformation is more reliable than its testimony about the period after it. Each paper names this limit at the relevant point: an “unresolved question” in the record, a “hypothesis that has not been coded” in the standard, a “hypothesis put forward” in counting, and a “proof of possibility” in claims.
Nor do they claim that the waqf (a legal-institutional arrangement in which property is dedicated to a designated purpose under a founder’s stipulations and managed by an administrator) has a unique structure of entitlement. All four papers carried out the replacement test using a German purpose foundation under the German Civil Code (BGB), and in most of them the result was commonality rather than difference. This is a contribution to the dialogue between East and West, not a failure of that dialogue.
Where the Inquiry Will Be Taken Further
This line of inquiry is to be developed further in a separate synthetic study bringing together the two series, entitled “From the Production of Benefit to Entitlement”. Its place is outside this book because it serves both series; attaching it to either would misplace it.
The entitlement corpus is itself described in a separate resource paper, and the record of its construction and freezing is set out in a deposit file. Both are outside this book, and readers are referred to them.
The Programme’s Common Appendix
The four papers promised this appendix without including it. It is reproduced here in full, once.
Inference Protocol · Variable Matrix · Declaration of Dependence
This appendix is reproduced in full in the standalone versions of the four papers—“When the Reference Point of the Answer Changed”, “The Entry Standard: Measured Content, a Hypothesised Source”, “Four Counts with No Common Denominator”, and “Did Charity Become a Right?”—and appears once in the collected volume. It follows the list of sources and precedes each paper’s own appendix. A single version is used because all four papers draw on the same material: they must therefore follow the same rule of inference, and the evidence they share must be declared before their agreement is read as mutual corroboration.
Bandar — with the assistance of al-Warrāq · 1448 AH / 2026
① The Shared Inference Protocol
1.1 · The Evidentiary Ceiling of the Source, Which Governs the Entire Series
The material of the four papers is the Waqf Entitlement Corpus, approved release WEC-432/1.0. This is the record of an authority engaged in framing. The muftī (jurist who issues fatwās) states the rule and legally characterises the facts; he does not control access, disburse revenue, adjudicate disputes or maintain a statistical register. It follows that this corpus is neither an administrative record, nor an implementation record, nor a central statistical enumeration; neither its size nor the completeness of its coverage can make up for that.
Two constraints follow from this ceiling and apply to every figure and judgement in the series:
The first is a constraint on the layer of analysis. Four layers must be kept distinct: the actor’s motivation, the design of the institution, its operation, and its outcomes. The corpus provides strong evidence for the second, and for the first only to the extent that the questioner and the muftī explicitly state it; it provides evidence for the third and fourth only in a few instances whose outcomes are recorded in their texts. Approximately ninety-seven per cent (97%) of the instances in the corpus have no documented outcome.
The second is the access constraint. The record preserves only those whose questions reached Dār al-Iftāʾ al-Miṣriyya (the Egyptian fatwā authority). It therefore omits blocked avenues of access, silent groups, and awqāf (plural of waqf) in which record-keeping was neglected. The direction of this bias is known in all four papers: it presents the old waqf system in a more favourable light and exaggerates the difference between it and what followed. The difference presented in any comparison in the series is thus an upper bound, not an estimate, and any reading that treats these papers’ proportions as rates for the population of entitlement-holders asks more of the source than it can support.
1.2 · The Three Evidentiary Tiers
Every claim in the four papers is assigned to one of three evidentiary tiers, and its wording must conform to its tier.
The first tier: what the fatwās establish. This comprises what is present in the text itself, once the editor’s layer has been separated from the question and answer: the formulation of the claim as submitted by the questioner; the instrument cited in its text; the basis of the muftī’s answer, whether Islamic jurisprudence with named sources, a founder’s stipulation, or a numbered statutory provision; the structure of the quoted stipulation and its constituent elements; counting terms and figures; and the muftī’s statement of the elements of a claim—standing, subject matter, forum and remedy. This tier permits the language of establishment, because its material consists of the text’s express words, not inferences drawn from it. Its limit is that it establishes what occurred in framing, not what occurred in disbursement.
The second tier: what the fatwās make likely. This comprises patterns and trends across decades, measured against an explicit denominator, after controlling the denominator and testing the declared falsifier. These include the trend in the shift of the reference point for framing, the density of standard statements, the distribution of rule types, and the share of instances involving observational counts. This tier permits expressions of likelihood, but not conclusive statements, because an alternative remains in each case: the measured change may be in what is published rather than in what occurs. This alternative has been tested collectively along four axes; the findings are in §1.4.
The third tier: what remains hypothetical and requires administrative archives or evidence from the courts. Every claim about the operational layer or outcomes belongs here. These include: the transfer of the beneficiary’s eligibility record to a file controlled by the apparatus; a change in the source of the rule preceding a change in its content; narrowing of the legal channel after 1959; the shift of the unit of counting to a programme category within the planning apparatus; and success rates, regularity of disbursement and the effectiveness of challenges. Claims in this tier are presented as hypotheses with their falsifiers, and are not expressed as findings.
The specific sources that would allow claims in the third tier to be upgraded are known: the Ministry of Awqāf’s administrative records and nāẓir (trustee-administrator) reports; the collection of Council of State judgements between 1959 and 1971; the operational ledgers of riwāqs (residential halls) and individual awqāf; Dār al-Iftāʾ’s incoming register; the planning reports and the statistical digest in their original form; and printed editions of the collection against which to compare a sample. No claim can be upgraded from the third tier to the second without consulting one of these sources; accumulating indications within the same corpus cannot upgrade it.
1.3 · Rules of Citation and Counting
First, the release tag. The tag WEC-432/1.0 is given at the first citation in every paper; thereafter, the identifier F432-NNNNN suffices, accompanied by the location of the text, the muftī and its date. Published texts do not cite the two intermediate layers.
Second, the difference between 405 and 396 is stated explicitly rather than left implicit. The third layer comprises four hundred and five candidate instances; the fourth comprises three hundred and ninety-six, after nine were excluded by manual coding. These two counts remain liable to confusion across the papers, so each aggregate count must state the layer on which it was calculated.
Third, the editor’s layer is separated out in every published count. The record has three layers: a title supplied by the collection’s editor, then a block of principles constituting the editor’s summary, then the question and answer. Conflating them attributes a later editor’s wording to the muftī; the effect of separating them has been measured, not estimated.
Fourth, the text-length field is defective. The recorded field measures the text together with extraneous material from the publishing website, amounting to an almost constant four thousand eight hundred and ninety-three characters; this compresses relative differences between decades to roughly a third of their size. Every text-length figure in the series is calculated using only the cleaned text.
Fifth, the derived layer must be cleaned before reuse. The two files of standard instances were built from the third layer rather than the approved release; they contain thirteen rows from six excluded records, accounting for approximately 0.85% of the rows.
Sixth, the source’s silence is recorded as “unknown”. It is treated neither as evidence for the thesis nor as evidence that negates it.
1.4 · Findings of the Collective Test of the Competing Explanation, and Their Limits
The explanation that the series’ trends measure bias in the corpus rather than structural transformation was tested collectively. The test covered four axes: the length of the main text, the composition of issue types, the identity of the questioner, and publication density and the representation of periods.
The judgement is that the competing explanation is partially ruled out. It is ruled out for the series’ principal trend, the shift of the reference point for framing from Islamic jurisprudence to codified law: the combined contribution of the four axes to explaining its 72.9-point gap is virtually zero, and the gap actually widens after normalisation. Three of its factual premises are correct and established: the text became longer, the composition of issues changed, and the identity of the questioner changed. Yet the conclusion does not follow from them, because these changes occur in a direction that does not generate the trend, and some run in the opposite direction.
The competing explanation remains viable in two places, which must be explicitly identified at the relevant points in the papers:
The first is the tail after 1959. Representation in the final two decades falls to 0.37 and 0.19, with a sample of seventeen texts followed by four. Every claim covering this period must therefore be downgraded to a description of the sample, or the series must end at 1959.
The second is the magnitude of the trend in entry standards in the second paper of this series. The rise in lineage statements in the 1950s may reflect a change in the content of incoming questions, or it may reflect editorial selection. The corpus cannot distinguish between them from within; doing so requires Dār al-Iftāʾ’s incoming register, comparison of editions, or the archives of the partition committees.
② Variable Matrix
This table is intended to prevent “category”, “unit of counting”, “rights-holder” and “claimability” from becoming four names for a single institutional transformation. For each paper, it therefore presents the dependent variable, the falsifier and the result of testing it, and the evidence that distinguishes the paper from its companions.
| Paper | Dependent Variable | Falsifier and Result | Distinct Evidence Not Shared with the Other Papers |
|---|---|---|---|
| 1 · When the Reference Point of the Answer Changed | Mode of inference in the muftī’s answer, seven categories, by decade | (a) The effect of the dating key or of a single muftī: tested using two keys and controlling for the muftī, and ruled out. (b) The replacement hypothesis—the document recedes in proportion to the rise of law: tested with two measures and ruled out | A field recording the mode of inference, manually coded for 396 texts · the counter for invocation of documents, controlled for length · the questioner field · the case record of the al-Minshāwī waqf · thirteen instances in which the document was absent · five instances of registration |
| 2 · The Entry Standard: Measured Content, a Hypothesised Source | Content of the standard statement, seven categories, and rule type, five categories, by decade | (a) Whether the density of lineage statements in the 1950s remains above the highest level in earlier decades after excluding liquidation: the specified test condition was met in 10 of the 12 measurement combinations, ruling out the tested quantitative version of the translation explanation. (b) A steady decline in the share of statements with multiple contents: not observed | 1,525 standard statements extracted from 348 texts · a content map covering 696 statements whose content was classified · the five-way distinction between rule types · the juristic apparatus of the standard: the wealth threshold, the attribution rule for maintenance, and the burden of proof · a negative case in which a standard ceased to operate half a century before the law |
| 3 · Four Counts with No Common Denominator | The unit of counting the receiving side: the object of counting, its unit and its scope of aggregation | The presence of a figure counting recipients as an observed population: met in a restricted form, as the numerical survey identified 17 records verifying actual presence, whose unit was the designated name and whose scope was an individual waqf. No aggregate figure above the level of the individual waqf was found | The three lexical and numerical surveys · the absence of the term “beneficiary” (mustafīd) across all nine decades · al-Azhar’s 1875 enumeration and its arithmetic audit · the Ḥanafī juristic discussion of the “threshold of enumerability”, with its sources · detailed statements of account |
| 4 · Did Charity Become a Right? | The structure of the claims bundle along five textual dimensions: standing, subject matter, forum, scope of review and remedy | Five separate falsifiers, H1–H5, each sufficient on its own, all outside the material examined by the paper. Two tests were implemented: cessation of the corpus after 1959 and the occurrence of a blocked route within it; neither was found | 27 instances selected along the dimensions of the bundle · ten enumerated instances of rejection, with their grounds · the course of a single case from the committee’s decision in 1960 to the Supreme Administrative Court’s judgement in 1971 · the practice indicator: cost, time and stay of execution · evidence from Khayrī Ḥammād, 1964 |
2.1 · Where Two Papers Measure the Same Thing
The first overlap concerns the first and second papers of this series on the source axis: one measurement, not two. The second paper of this series separates the content axis from the axis of institutional source, but does not code the latter at the level of statements or produce a single figure for it. Its account of the shift in the source of the rule rests on examined instances and quoted texts; its quantitative support is precisely the first paper’s count of modes of inference. It follows that, on this axis, the two papers measure the same variable with the same instrument, and that the content axis alone is distinctive to the second. This is detailed in §③.
The second overlap concerns the second and fourth papers of this series in their use of statutory texts: one source, two axes. Both papers examine the laws between 1946 and 1971. The second, however, reads them to identify who sets the entry rule and who interprets it, while the fourth reads them for the elements of the bundle: before whom an action is brought, concerning what, and with what remedy. The axes are distinct, but the source is the same. Their agreement that 1946 marks a turning point therefore constitutes two readings of a single text, not two independent pieces of evidence.
The third overlap concerns the first and third papers of this series, which examine the same text using different units. The first counts the instance and identifies the reference point of its answer; the third counts words and figures in the text. Their units therefore do not coincide, and their independence lies in the unit and the instrument, not in the source.
The overarching rule across these three overlaps is that the four papers draw on a single corpus. Agreement between two of them must therefore not be read as independent corroboration, nor must the recurrence of a finding across them be counted as independent replication. Independent corroboration in this series requires a source outside the corpus, from among those named in §1.2.
③ Declaration of Dependence Between the First and Second Papers of This Series
The following text is reproduced verbatim in both papers—“When the Reference Point of the Answer Changed” and “The Entry Standard: Measured Content, a Hypothesised Source”—at the point where each states its contribution, with a reference to this appendix.
Declaration of dependence. This paper and the other paper named above are two analyses dependent on a single evidentiary base, rather than two independent studies whose findings agree. They attribute the same transformation to the same texts in the Waqf Entitlement Corpus, release WEC-432/1.0. “The Entry Standard: Measured Content, a Hypothesised Source” has no quantitative coding of the institutional-source axis: this axis was not coded at statement level and produced no numerical result in the paper. Its quantitative support for the claim that the source shifted is therefore the counter for the mode of inference implemented in “When the Reference Point of the Answer Changed”. It follows that the test of the shared source axis occurs only in the first paper of this series, while the second is confined to the structure and content of the standard and the mechanisms of its abolition, translation and recombination. Agreement between the two papers consequently does not constitute independent corroboration: the evidence, the reader and the measurement on the shared axis are all the same. Independent corroboration requires a source outside this corpus: the incoming register of Dār al-Iftāʾ al-Miṣriyya (the Egyptian fatwā authority), the Ministry of Awqāf’s administrative records or comparison with the printed editions.
3.1 · Editorial Implications of the Declaration
Four editorial measures follow from this declaration:
First: the second paper of this series refers to the first as the location of the test, not as evidence corroborating the claim about source. The wording of the reference is “the test is in the first paper of this series”, not “this is also supported by the first paper of this series”.
Second: the second paper of this series makes no claim about temporal precedence between changes in source and changes in content. Content is measured in a series organised by decade, while source is described through texts dated by the events they record; what has been measured cannot be ordered in time relative to what has not been measured.
Third: at the point where it presents its finding, the first paper of this series states that the account built upon it in the second draws on the first paper’s own count, so consistency between the two is not read as independent verification.
Fourth: both papers state that the sample-based assessment of coding reliability between two readers has not yet been conducted, and that neither the manual coding in the first nor the automated coding in the second has a calculated agreement rate. There is one reader on both sides of the dependence, adding a third constraint to those of source and instrument.
Bandar — with the assistance of al-Warrāq.