A narrower difference in the catalogue, not a verdict on publishing
Across the 72 subject and form classes reported by the Bibliothèque nationale de France, the composition of notices flagged as self-published was closer to the composition of all notices in bibliography year 2024 than in 2023. The total-variation distance fell from 34.0713 to 31.3490 points on a zero-to-100 scale, a decrease of 2.7222 points. This result describes two annual outputs of the Bibliographie nationale française – Livres. It does not show that books published or deposited in those years became more alike, because the year a notice appears in the bibliography is not necessarily the year its book was published or deposited.
The distinction changes the question that can responsibly be answered. These tables support a comparison of recorded catalogue composition: which subjects and forms account for the notices processed into each annual bibliography, and how the flagged subset differs from the whole. They do not measure the complete self-publishing market or the distribution of new works produced during a calendar year. Within this narrower scope, the result is more informative than a fiction-versus-documentary comparison. It incorporates every available classification code, identifies changes moving in opposite directions, and tests whether the inclusive benchmark itself explains the apparent narrowing. Both the complete classification and a broader grouping show a decline, but neither establishes a lasting trend beyond the two observed cohorts.
What is counted, and when it is counted
The 2023 bibliography-year tables contain 77,541 notices, of which 16,311 carry the official self-publishing flag. The corresponding 2024 figures are 70,481 and 11,246. Subtracting the flagged counts gives unflagged remainders of 61,230 and 59,235 notices. These are administrative record counts. They are not numbers of copies printed or sold, numbers of authors, or conservatively matched unique intellectual works. Nor should different editions and formats be collapsed into a single title without a separate, documented matching procedure. This analysis retains the source's notice unit rather than imposing a more ambitious interpretation on it.
The dates are equally specific. The BnF documentation defines the annual bibliographic extraction through the year in which notices appear in the national bibliography. Deposit registration and cataloguing are different stages, and processing can carry material from one calendar year into another. Consequently, a smaller flagged cohort in 2024 cannot by itself establish a fall in self-published book output. It could reflect a different combination of received material, processing timing and classification coverage. The numerical comparisons here remain valid descriptions of the tables, but the mechanism producing those tables is not observed. A genuine deposit-year study would need compatible deposition dates and sufficiently complete later cataloguing, not a substitution of registration totals for the notice denominators used here.
An official flag, not a publisher-name guess
Self-publishing status is taken from an explicit BnF catalogue criterion, not inferred by comparing an author's name with a publisher's name. The documented extraction uses positions 17 and 18 of the Intermarc Guide 000 field, with values 1 and h respectively. This makes the comparison reproducible as a comparison of recorded flags. It does not turn the flag into an independently validated account of every book's financing, contractual arrangements or publishing history. The scope associated with self-publishing in the BnF material encompasses more than an author personally performing every publishing task; author-funded editions and service-provider arrangements also matter.
For that reason, the complementary group is consistently called the unflagged remainder, or notices not recorded as self-published. It is not a verified category of traditional publishers, and it should not be described as such. A difference between the two groups may depend partly on recording practices or incomplete identification. The source also identifies a relatively recent introduction of the catalogue distinction and warns about processing delays affecting self-published material. Extending these results backwards as though the flag had always operated identically would therefore be unjustified. The purpose of using the official flag is to avoid an invented classification, while keeping the limits of the actual classification visible.
Subjects and forms are not all literary genres
The principal comparison uses the BnF's three-digit cadre de classement. This classification combines subjects and forms across the catalogue. It includes, for example, code 340 Droit and code 610 Sciences médicales et paramédicales as well as code 801 Poésie and code 803 Romans et fiction romanesque. Calling all 72 categories literary genres would obscure this distinction. The exact French code labels are retained in the evidence table so that an English explanation does not silently replace the source taxonomy. All 72 labels matched across the two workbooks, allowing the comparison to use the same named classification rather than a reconstructed crosswalk.
A separate source field distinguishes documentary, fiction, mixed and blank entries. That field answers a different, coarser question and is not used to create the 72-class distance. In particular, the subject/form analysis does not assume that every class within the broad literature domain is interchangeable with the fiction flag. Keeping these dimensions separate prevents a familiar headline about fiction from standing in for the whole distribution. It also makes the result interpretable: any contribution to the distance belongs to a named subject or form, and the full table shows the signed share difference behind it. The measure is not a score of literary merit, thematic richness within individual books, or variety experienced by readers.
How the full-distribution distance works
For each class, the calculation divides its flagged-notice count by the total flagged count and its overall count by the overall notice total. It takes the absolute difference between these two shares, adds the differences across all 72 classes, and divides the sum by two. This is total-variation distance. The division by two avoids counting both sides of the same redistribution: a larger share somewhere necessarily accompanies smaller shares elsewhere when both distributions sum to one. A value of zero means identical class shares. On the displayed zero-to-100 scale, a value of 100 would mean that the two distributions occupy entirely different classes.
The reported 34.0713 and 31.3490 points are therefore distances between complete distributions, not the prevalence of a particular subject. Their 2.7222-point difference is a change in that distance, not a percentage change in book production. Exact counts are used throughout. The unscaled inclusive distance is 143641201/421590417 in 2023 and 124240806/396314663 in 2024; displayed decimals are rounded versions of these rational values. The exact fractions matter for reproducibility, not because the administrative classification is perfectly certain. Computational precision and substantive coverage are different questions. No sampling confidence interval is attached to these administrative totals, and numerical exactness does not remove uncertainty about missing deposits, flagging or processing delays.
Why comparison with all notices needs a correction in interpretation
The all-notice benchmark includes the flagged notices themselves. This is useful for describing how a visible subset differs from the catalogue as a whole, but it automatically makes the two distributions look closer than a comparison with the remainder would. Let w denote the flagged share of all notices. The overall distribution equals w times the flagged distribution plus (1 − w) times the unflagged distribution. It follows exactly that the distance between flagged and all notices equals (1 − w) times the distance between flagged and unflagged notices. This is a property of the inclusive benchmark, not an empirical discovery about publishing behaviour.
The flagged share falls from 21.0353% of notices in 2023 to 15.9561% in 2024. The attenuation therefore changes between years. If the flagged-versus-remainder distance were unchanged, the lower flagged share would make the inclusive distance larger, because the whole would contain less of the flagged distribution. Here the observed inclusive distance instead declines. To make that result transparent, the analysis also calculates the exact remainder within every class. These counts are all nonnegative and exhaust the overall total after the flagged subset is removed. The comparison does not require guessing which unflagged notices belong to a commercial publisher or a particular business model.
The narrowing survives the unflagged-remainder comparison
Across the full 72-class scheme, the distance between flagged notices and the unflagged remainder falls from 43.1475 points in 2023 to 37.3008 in 2024. Its decrease of 5.8467 points is larger than the decrease against the inclusive benchmark. These are not competing estimates of one unidentified quantity: they answer two explicitly different questions. One asks how the flagged subset differs from the whole bibliography; the other asks how it differs from what remains after that subset is removed. Presenting both avoids attributing the changing inclusive distance entirely to the changing flagged share.
The mixture identity was checked with exact rational arithmetic in each year. It reconciles the two sets of results rather than relying on an approximate verbal explanation. Nevertheless, passing that check does not validate the completeness of the flag or explain why the underlying distributions changed. The observed remainder may contain material that would be classified differently with fuller information. The result can therefore be stated firmly but narrowly: the recorded subject/form composition of the flagged cohort is closer to the recorded composition of the unflagged remainder in 2024 than in 2023. It cannot establish convergence between two exhaustively identified publishing sectors, and two annual observations cannot establish the direction of subsequent change.
The largest differences do not all move together
Code 803 Romans et fiction romanesque has a much larger share among flagged notices than among all notices in both years. Its signed flagged-minus-all gap declines from 22.0795 percentage points in 2023 to 20.3952 in 2024. Code 801 Poésie also has a larger flagged share, with its gap declining from 6.5423 to 4.9952 points. Each class contributes half its absolute share gap to total variation. The changes in these two contributions are therefore minus 0.8422 and minus 0.7735 points respectively. This decomposition shows where part of the aggregate narrowing occurs without making either class a substitute for the full calculation.
Code 805 Bandes dessinées moves in the opposite direction. Its flagged share is lower than its overall share, and the signed gap changes from minus 5.4157 to minus 6.7015 points. Its contribution to the inclusive distance consequently increases by 0.6429 points. This lower share is solely arithmetic relative to the chosen benchmark. It does not mean that too few such books exist, that demand is unmet, or that a publishing model performs inadequately. The example matters because an overall decline can coexist with widening differences in particular classes. The complete evidence retains every class, including those whose contributions are small, rather than presenting selected movements as though they constituted the whole distribution.
What changes when the classification is made broader
The predeclared robustness comparison groups the three-digit codes by their first digit into ten broad domains. At that resolution, the inclusive distance declines from 22.5904 points in 2023 to 19.3495 in 2024, a decrease of 3.2409 points. Against the unflagged remainder, it declines from 28.6082 to 23.0230 points, a decrease of 5.5852. Thus the direction of the result is not confined to the finest available set of classes. Both the full classification and the broad grouping show smaller recorded composition differences in the later bibliography-year cohort.
The smaller levels at the broad resolution should not be read as a contradiction. Grouping can cancel differences between classes inside the same domain. For example, opposite share gaps among forms within a broad literature category can partly disappear when those forms are added together. The ten-domain result therefore answers a less detailed question, not a more accurate one. Its role is to show how sensitive the description is to classification resolution. Neither resolution is treated as a universal measure of diversity. A study interested only in differences between broad subjects would lose information that matters to a study of specific forms; a study using all 72 classes must acknowledge that the official scheme determines which distinctions can be observed.
Checks that make the comparison reproducible
The class tables were checked before the distance calculations. Their sums match all four relevant notice denominators exactly. The flagged table contains 68 represented classes in 2023 and 70 in 2024, while the overall table contains 72 in each year. Every represented flagged code is present in the corresponding overall table, and no flagged class count exceeds its overall count. Because the nonnegative flagged cells exhaust the official flagged total, omitted classes can be filled with zero for this aggregate comparison. That is a conclusion from these reconciled tables, not a general rule that missing catalogue fields or omitted source cells always mean zero.
A separate implementation then extracted the raw workbook tables and recalculated the distances using overlapping probability mass rather than repeating the primary sum of absolute differences. It matched the exact rational results, the code labels, the totals and the inclusive-mixture identity. The downloadable analytical evidence can consequently be kept compact: class counts and labels, derived differences, the calculation script and source references. Reproducing the result does not require distributing a list of individual books or the self-publishing metadata sheet containing thousands of titles. Independent arithmetic verification guards against extraction and calculation mistakes, while the separate discussion of units, cohorts and flag coverage guards against a different error: attaching an unsupported interpretation to a correct number.
What this adds, and what remains unanswered
The BnF has already reported that the self-publishing-flagged catalogue is more fiction-heavy than the overall catalogue. Repeating that contrast would contribute little. The additional analysis here is the complete subject/form distance, its decomposition across classes, and the demonstration that its direction survives both an unflagged-remainder comparison and a broader classification. It also makes the inclusive benchmark's mathematical attenuation explicit. Together, these elements replace a single binary contrast with a fuller description of the recorded distributions, including a case in which a particular difference widens while the overall distance declines.
The limits remain central to the answer. The data do not reveal whether the same pattern would appear after following identical deposit-year cohorts through complete cataloguing, nor whether changes in recording or processing contributed to the observed difference. They do not measure ebook-only activity, reader choices, sales or author income. No claim about improved or reduced publishing diversity follows from the smaller distance: becoming closer to a benchmark is not inherently better. The supported conclusion is that two administrative bibliography-year cohorts show a narrower subject/form composition gap under several transparent comparisons. Establishing anything broader would require additional evidence about cohort coverage and timing, not stronger language applied to these same tables.
Catalogue-composition distance decreases under both comparisons
France · 2023–2024 bibliography output · Complete official subject/form classification
Full chart and source
Scroll horizontally to inspect the chart. ↔
Chart and values
Scroll horizontally to see all columns.
| Classification | Comparison | 2023 (pp) | 2024 (pp) | Change (pp) |
|---|---|---|---|---|
| 72 subject/form classes | Flagged versus all | 34.0713 | 31.3490 | −2.7222 |
| 72 subject/form classes | Flagged versus unflagged remainder | 43.1475 | 37.3008 | −5.8467 |
| Ten broad domains | Flagged versus all | 22.5904 | 19.3495 | −3.2409 |
| Ten broad domains | Flagged versus unflagged remainder | 28.6082 | 23.0230 | −5.5852 |
Total variation multiplied by 100: a 0–100 percentage-point distance scale. All is an inclusive comparator. Unflagged remainder is a catalogue-flag complement, not verified traditional publishing. No confidence intervals.
Data & downloads
Scroll horizontally to see all columns.
| Classification | Comparison | 2023 (pp) | 2024 (pp) | Change (pp) |
|---|---|---|---|---|
| 72 subject/form classes | Flagged versus all | 34.0713 | 31.3490 | −2.7222 |
| 72 subject/form classes | Flagged versus unflagged remainder | 43.1475 | 37.3008 | −5.8467 |
| Ten broad domains | Flagged versus all | 22.5904 | 19.3495 | −3.2409 |
| Ten broad domains | Flagged versus unflagged remainder | 28.6082 | 23.0230 | −5.5852 |
Total variation multiplied by 100: a 0–100 percentage-point distance scale. All is an inclusive comparator. Unflagged remainder is a catalogue-flag complement, not verified traditional publishing. No confidence intervals.
Scroll horizontally to see all columns.
| Code | Official subject/form label | Flagged 2023 | All 2023 | Flagged 2024 | All 2024 |
|---|---|---|---|---|---|
| 000 | Généralités | 1 | 46 | 1 | 47 |
| 004 | Informatique | 23 | 344 | 18 | 229 |
| 010 | Bibliographie | 0 | 10 | 2 | 15 |
| 020 | Sciences de l'information | 2 | 38 | 1 | 39 |
| 030 | Encyclopédies générales | 0 | 36 | 0 | 22 |
| 060 | Organisations à vocation savante | 0 | 20 | 1 | 17 |
| 070 | Presse et édition | 6 | 76 | 7 | 64 |
| 090 | Bibliophilie | 2 | 46 | 1 | 34 |
| 100 | Philosophie | 129 | 1,091 | 87 | 1,006 |
| 130 | Esotérisme, phénomènes paranormaux | 287 | 1,066 | 250 | 1,029 |
| 150 | Psychologie, psychanalyse | 237 | 1,298 | 253 | 1,238 |
| 200 | Religion (ouvrages généraux et comparatifs) | 38 | 133 | 21 | 111 |
| 220 | Bible, judaïsme, christianisme | 271 | 1,866 | 179 | 1,676 |
| 290 | Islam, religions orientales et anciennes, autres religions | 50 | 377 | 31 | 378 |
| 300 | Sciences sociales | 123 | 1,799 | 91 | 1,408 |
| 320 | Sciences politiques | 217 | 1,454 | 117 | 1,396 |
| 330 | Economie | 129 | 1,195 | 74 | 1,010 |
| 340 | Droit | 35 | 1,887 | 33 | 1,797 |
| 350 | Administration publique | 11 | 272 | 8 | 255 |
| 355 | Art et science militaires | 20 | 203 | 6 | 241 |
| 360 | Problèmes et services sociaux | 85 | 1,219 | 60 | 1,065 |
| 370 | Education et enseignement | 58 | 694 | 38 | 589 |
| 380 | Commerce, communication, transports | 19 | 189 | 18 | 184 |
| 390 | Ethnologie, anthropologie, folklore | 23 | 249 | 11 | 217 |
| 400 | Langues et linguistique | 78 | 860 | 125 | 802 |
| 500 | Sciences (généralités) | 11 | 136 | 2 | 86 |
| 510 | Mathématiques | 23 | 253 | 13 | 193 |
| 520 | Astronomie | 7 | 112 | 9 | 115 |
| 530 | Physique | 11 | 163 | 17 | 134 |
| 540 | Chimie, minéralogie, cristallographie | 2 | 49 | 0 | 26 |
| 550 | Sciences de la Terre | 8 | 111 | 8 | 113 |
| 560 | Paléontologie | 2 | 46 | 1 | 86 |
| 570 | Sciences de la vie | 10 | 304 | 10 | 255 |
| 580 | Botanique | 17 | 198 | 9 | 148 |
| 590 | Zoologie | 39 | 467 | 24 | 386 |
| 600 | Sciences appliquées | 0 | 27 | 1 | 18 |
| 610 | Sciences médicales et paramédicales | 164 | 2,351 | 91 | 1,738 |
| 620 | Sciences de l'ingénieur | 46 | 627 | 43 | 540 |
| 630 | Agriculture, élevage | 47 | 534 | 30 | 437 |
| 640 | Economie domestique, hôtellerie | 108 | 1,678 | 133 | 1,336 |
| 650 | Gestion et organisation de l'entreprise | 96 | 1,150 | 76 | 951 |
| 660 | Génie chimique et technologies connexes | 7 | 83 | 5 | 70 |
| 670 | Industries | 2 | 18 | 1 | 14 |
| 680 | Métiers divers | 7 | 53 | 6 | 35 |
| 690 | Bâtiment | 2 | 91 | 2 | 68 |
| 700 | Arts | 41 | 1,326 | 30 | 1,050 |
| 710 | Urbanisme | 3 | 102 | 7 | 113 |
| 720 | Architecture | 28 | 425 | 14 | 442 |
| 730 | Arts plastiques, sculpture | 23 | 240 | 15 | 248 |
| 740 | Dessin, arts décoratifs, artisanat d'art | 88 | 978 | 70 | 1,006 |
| 750 | Peinture | 51 | 655 | 35 | 588 |
| 760 | Arts graphiques | 9 | 73 | 25 | 94 |
| 770 | Photographie | 117 | 841 | 121 | 768 |
| 780 | Musique | 21 | 506 | 19 | 462 |
| 790 | Spectacles et loisirs (généralités) | 10 | 45 | 7 | 54 |
| 791 | Audiovisuel | 20 | 410 | 9 | 378 |
| 792 | Représentations scéniques | 9 | 138 | 3 | 118 |
| 793 | Jeux | 50 | 569 | 63 | 677 |
| 796 | Sports | 67 | 504 | 63 | 607 |
| 800 | Histoire et critique littéraires | 65 | 1,288 | 31 | 1,127 |
| 801 | Poésie | 1,841 | 3,679 | 1,028 | 2,922 |
| 802 | Théâtre | 239 | 809 | 124 | 719 |
| 803 | Romans et fiction romanesque | 7,397 | 18,044 | 4,924 | 16,485 |
| 804 | Essais littéraires et écrits divers | 723 | 1,799 | 436 | 1,345 |
| 805 | Bandes dessinées | 87 | 4,613 | 85 | 5,256 |
| 809 | Littérature pour la jeunesse | 991 | 6,251 | 869 | 6,371 |
| 900 | Histoire et géographie (généralités) | 13 | 113 | 9 | 106 |
| 910 | Géographie, tourisme (guides) et voyages | 103 | 1,245 | 110 | 997 |
| 920 | Biographies et généalogie | 1,606 | 5,723 | 1,047 | 4,779 |
| 930 | Préhistoire et histoire ancienne | 27 | 313 | 3 | 298 |
| 940 | Histoire médiévale, moderne et contemporaine (sauf la France) | 55 | 890 | 32 | 800 |
| 944 | Histoire de la France (depuis 486) | 174 | 1,043 | 153 | 1,053 |
All 72 codes; original French labels retained. Counts are bibliography-year notices, not books published or deposited that year. Zero-filled flagged categories are justified by exhaustive reconciled source totals. All includes flagged.
Scroll horizontally to see all columns.
| Bibliography year | Flagged notices | All notices | Unflagged remainder | Flagged share (%) |
|---|---|---|---|---|
| 2023 | 16,311 | 77,541 | 61,230 | 21.0353 |
| 2024 | 11,246 | 70,481 | 59,235 | 15.9561 |
Flagged share is a share of annual bibliography output, not a self-publishing market share.
Evidence & method
How we calculated it
- Descriptive comparison of all 72 aligned BnF cadre de classement codes in the 2023 and 2024 bibliographic books workbooks. Exact French subject/form labels retained.
- Use official autoédition catalogue flag: Intermarc Guide 000 position 17=1 and 18=h. All notices are an inclusive comparator; the exact unflagged remainder is computed by subtraction, not inferred from names.
- TV is one half of the sum of absolute class-share differences. Multiply by 100 for the reported distance scale. Primary outcome is 2024 minus 2023 distance; exact rational values accompany displayed decimals.
- Predeclared sensitivities: ten first-digit subject domains and comparison with the unflagged remainder. Verify TV(flagged, all) = (1 − flagged share) × TV(flagged, remainder).
- All four class totals reconcile with official notice denominators; omitted flagged cells are zero only within exhaustive reconciled tables. Independent raw-XLS extraction and overlap-mass arithmetic reproduce the results.
What the data cannot tell us
- Bibliography year reflects processing output, not deposit or book publication year. Differential delays and flag coverage can change cohort composition.
- A notice is not necessarily a unique title or intellectual work, and is not a sales copy. Entry and bibliography units can differ.
- The flag includes a broad autoédition circuit; the remainder is not established traditional or non-self-published output. No inference from author/publisher name similarity.
- Dewey-derived subject/form classes mix topics, forms and audiences; they are not pure literary genres. Classification resolution affects distance.
- No claims about market shares, sales, demand, income, quality, missing-deposit rates, ebook-only activity or national production diversity. Two catalogue years do not establish a long-run or causal trend.
- Administrative totals do not justify sampling confidence intervals or population coverage claims. Displayed precision is arithmetic, not measurement certainty.
Original sources
- BnF 2023 bibliographic books workbook
BnF 2023 bibliographic books workbook: Bibliography-publication year 2023; complete cadre de classement counts in Autoédition and Répartition thématique. Exact workbook labels preserved; annual entries and book publication dates are not substituted.
Accessed October 30, 2026
- BnF 2024 bibliographic books workbook
BnF 2024 bibliographic books workbook: Bibliography-publication year 2024; complete cadre de classement counts in Autoédition and Répartition thématique. Exact workbook labels preserved; annual entries and book publication dates are not substituted.
Accessed October 30, 2026
- BnF reservoir documentation, 2023–2024
BnF reservoir documentation, 2023–2024: Pages 7 and 10–12 distinguish entries from bibliography output; classification uses Intermarc 680$a; annual cohort 015$a; autoédition flag Guide 000 positions 17=1 and 18=h. Fiction is a separate 009-position-03 field.
Accessed October 30, 2026
- BnF dataset and reuse designation
BnF dataset and reuse designation: Producer Bibliothèque nationale de France; dataset metadata designates Licence Ouverte / Open Licence (identifier fr-lo). This designation applies to the source dataset, not an invented licence for the article.
Accessed October 30, 2026
- BnF Observatoire archive
BnF Observatoire archive: Lists 2023/2024 reservoir releases and describes reuse of these public data as free; distinguishes these releases from the 2022 synthesis.
Accessed October 30, 2026
- BnF 2022 synthesis
BnF 2022 synthesis: Printed pages 19–23 describe catalogue delays, especially for autoédition, and the criterion introduced at end-2020. Already publishes the rounded 46% overall versus 64% autoédition fiction contrast; that contrast is background, not this study’s original result.
Accessed October 30, 2026
- BnF bibliography content and cataloguing rules
BnF bibliography content and cataloguing rules: Official indexed explanation of abbreviated autoédition records and broad author-publishing/author-funded/service-provider scope. Bibliography output is an activity cohort; multi-volume books and collection treatment can produce different entry and notice counts.
Accessed October 30, 2026
Cite & reuse this study
Alisson Bento. Self-publishing flags and France’s catalogue mix, 2023–2024. Secondary analysis of Bibliothèque nationale de France, Observatoire du dépôt légal, bibliographic books indicators for 2023 and 2024. Sources accessed October 30, 2026. https://alissonbento.com/research/france-selfpublishing-catalogue-composition-2023-2024/
Source data: Bibliothèque nationale de France, Observatoire du dépôt légal, bibliographic books indicators for 2023 and 2024. Official dataset: https://www.data.gouv.fr/datasets/observatoire-du-depot-legal. Accessed 30 October 2026. Dataset metadata identifies Licence Ouverte / Open Licence, identifier fr-lo. The BnF archive also states that reuse of these public data is free: https://www.bnf.fr/fr/observatoire-du-depot-legal-editions-des-annees-precedentes. No particular licence version is invented from the identifier. Extraction, calculated aggregates, English explanation and original figure: Alisson Bento. Changes consist of selecting the complete subject/form count tables, aligning codes, deriving unflagged complements and distribution distances, and rendering an English presentation. Preserve BnF attribution and the source/version references when reusing source-derived aggregates. No BnF endorsement or logo rights are claimed. This source designation is not a declaration of a licence for the entire article, third-party materials, or the DEPS report. Only compact selected aggregates and references are distributed, not complete source workbooks or bibliographic record lists.
Corrections or editorial questions