Reviewer3 (AI review)

September 26, 2026

AI Review (reviewer3.com)

Comments (8)

1. Inconsistent Chinese characters for Yan Shu — Editorial

The image of “ the shade of the willows, ” an image never mentioned at all in the Xiangyang shoucheng lu , was inserted into the Bi'wo ji — very likely because “ the shade of the willows ” recurs repeatedly in poetry, most notably in a line by Yan Shu 晏殊 , a poet who exerted considerable influence on the literati of the Song

The Chinese characters used for the name of the poet Yan Shu (and his son Yan Jidao) are inconsistent. In the main text on page 9, his name is written as 'Yan Shu 晏殊', but in footnote 22 on the same page, it is written as 'Yan Shu 晟殊' and 'Yan Jidao 晟幾道' (using the character '晟' instead of '晏').

2. Inconsistent Chinese characters for Bi'wo ji — Editorial

Zhao

The Chinese characters used for the title of the work 'Bi'wo ji' are inconsistent. In the abstract on page 1, it is written as '裨幄集' (where '幄' correctly means tent, matching the translation 'Collection from the Staff Tent'). However, in footnote 13 on page 4 and footnote 17 on page 5, it is written as '裨幣集' (using the character '幣', which means currency or money).

3. Inconsistent Chinese characters for Duan Xizhong — Editorial

The principle of being “ meticulous about campaigns of war ” has already been discussed by Duan Xizhong 段 熙 仲 , who offered a persuasive explanation of Dong Zhongshu's ordering of precedence in campaigns of

The Chinese characters used for the name of the scholar Duan Xizhong are inconsistent. In the main text on page 2, his name is written as '段 熙 仲' (Duan Xizhong), but in footnote 4 on the same page, it is written as '段 熊 仲' (Duan Xiongzhong).

4. Contradiction in which text is polished — Critical

2. Aside from a small number of spontaneously lyrical poems, the Xiangyang shoucheng lu shows clear signs of having been polished in comparison with the Bi'wo ji — a point that will be elaborated in detail below.

Point 2 on page 5 states that the 'Xiangyang shoucheng lu shows clear signs of having been polished in comparison with the Bi'wo ji.' This directly contradicts Point 1 on the same page, which states that 'the Bi'wo ji is considerably more embellished in its literary style.' It also contradicts the entire thesis of the paper, which argues that the Bi'wo ji is a polished, literary, and embellished version of the more colloquial and unadorned Xiangyang shoucheng lu.

5. Mischaracterization of sincerity piercing the sun — Major

The Shoucheng lu , written during the war, is relatively unadorned, whereas in the Bi'wo ji , composed after the war, Zhao Wannian polished the fairly direct laudatory expression “ losyal sincerity pierced the sun ” by tracing it back to the allusion of Zhang Xun 張巡 and Xu

The author characterizes the phrase 'loyal sincerity pierced the sun' (忠誠貫日) in the Shoucheng lu as a 'fairly direct laudatory expression' and contrasts it with the 'allusion' to Zhang Xun and Xu Yuan in the Bi'wo ji. However, 'sincerity piercing the sun' (cheng guan ri / dan xin guan ri) is itself a well-known classical allusion with deep roots in early Chinese literature (such as Zou Yang's Han-dynasty writings). Treating it as an unadorned, non-allusive expression weakens the contrast being made between the 'raw' prose and the 'allusive' poetry. The authors should acknowledge that both texts employ classical allusions, albeit of different types.

6. Misattributed narrative subject in fire account — Major

By inserting this highly evocative four-character expression, Zhao Wannian in effect sacrificed narrative accuracy, transforming an account in which the wind happened by chance to arise at the moment of setting the fire into one in which the Southern Song side deliberately used the wind to set the fire — reversing the logical sequence of events.

On page 9, the author argues that by using the phrase 'taking advantage of the wind, [they] set the fire loose' (yin feng zonghuo), Zhao Wannian 'transformed an account in which the wind happened by chance to arise at the moment of setting the fire into one in which the Southern Song side deliberately used the wind to set the fire.' However, in both the Shoucheng lu ('虜人縱火... 忽大風起') and the Bi'wo ji ('因風縱火'), the subject performing the action of setting the fire is the enemy (the Jin forces, '虜人'), not the Southern Song defenders. The author has misattributed the action to the Southern Song side, which undermines the argument about the reversal of logical sequence and deliberate strategy by the defenders. The authors should correct this reading of the narrative subjects.

7. Misunderstanding of military term niuwu — Major

The term “ niuwu ” 牛屋 in the Shoucheng lu is almost incomprehensible; Julie J. Avery renders it in English as “ ox stable.

The analysis of the term 'niuwu' (牛屋) on pages 7–8 claims that the term is 'almost incomprehensible' and that the Bi'wo ji's use of 'rhinoceros hide' (xige) explains or clarifies it. However, 'niuwu' (literally 'ox-shed' or 'ox-house') is a well-documented historical term in traditional Chinese military treatises (such as the Wujing zongyao) referring to a specific type of mobile protective shelter used during sieges. It is not an incomprehensible term or a mistake, and the 'rhinoceros hide' in the Bi'wo ji refers to the material covering such structures rather than serving as an explanation of the term itself. The authors should revise this discussion to reflect established military terminology.

8. Contradictory statement on text polishing — Major

Aside from a small number of spontaneously lyrical poems, the Xiangyang shoucheng lu shows clear signs of having been polished in comparison with the Bi'wo ji — a point that will be elaborated in detail below.

In the list of evidence establishing the relative chronology of the two works on page 5, the second point states that the 'Xiangyang shoucheng lu shows clear signs of having been polished in comparison with the Bi'wo ji.' This directly contradicts the central thesis of the paper, which argues that the Bi'wo ji is the later, polished, and embellished version of the earlier, more raw Xiangyang shoucheng lu. The authors should correct this sentence to ensure logical consistency with the rest of their analysis.

xPeerd (AI review)

September 26, 2026

AI Review (xPeerd — DAReview)

Summary

This manuscript examines how Chinese historians learned to write about war by comparing Zhao Wannian’s Xiangyang shoucheng lu and Bi’wo ji. Its central claim is that a broader “culture without soldiers” shaped historiographical practice such that later writing about war became increasingly literary, generalized, and mediated by inherited textual conventions rather than direct experiential description. The paper’s main strength is the close comparative reading of paired passages from the two texts, which gives the argument a concrete evidentiary basis. It also situates the case within larger historiographical debates through references to the Chunqiu tradition, Fairbank, Franke, and later scholarship.

The principal weaknesses are analytical overreach, uneven methodological explicitness, and insufficient control over alternative explanations. At several points, the paper moves from a compelling case-specific observation about genre transformation to broader claims about Chinese historiography as such. The manuscript also relies heavily on interpretive inference without always defining criteria for identifying “borrowing,” “imitation,” “generalization,” or “distortion.” In addition, the prose occasionally becomes rhetorically elevated or self-qualifying in ways that dilute argumentative precision. Overall, the manuscript appears intellectually serious and coherent, with no obvious signs of fabrication, but it would benefit from major revision to tighten its claims, sharpen its method, and better delimit the scope of its conclusions.

Potential Major Revisions

1. Clarify the article’s research question and distinguish it from its broader interpretive ambition.

The paper opens with a large conceptual claim: “historians would sometimes be unconsciously infected by earlier textual records and carry these records into their own writing” and that this produced “a kind of doubling, or ‘afterimage.’” This is provocative, but the manuscript does not early enough separate what is being demonstrated directly from the Zhao Wannian case and what is being hypothesized about Chinese historiography more broadly. The phrase “This article draws on an example from the Chinese historiographical tradition to explain such a shift in historical understanding and thinking” is too elastic and leaves unclear whether the paper is a tightly bounded case study or a general theory-building intervention.

Revision needed: state explicitly, near the end of the introduction, whether the article argues: a) that Zhao Wannian’s two texts exemplify one possible pattern of transformation in war writing, or b) that they reveal a general mechanism in premodern Chinese historiography.

At present, the conclusion leans strongly toward the latter, for example: “The culture without soldiers led traditional Chinese scholars and official historians, in narrating war, to have no choice but to seek out means of understanding war from pre-existing texts.” That formulation is too categorical unless the evidentiary base is widened or the claim is narrowed.

2. Define the method of comparison more rigorously.

The strongest part of the article is Section II, where the author compares concrete shifts between the two texts. For example, “The two watchtowers, more than ten large cannons, and several hundred ox-hide shelters were all demolished” becomes “[They] burned their own siege equipment and set fire to their camp.” Likewise, “the caitiff bandits built a pontoon bridge at Dongjin” becomes “caitiff cavalry choked Dongjin.” These examples are useful, but the article does not fully specify the principles by which changes are classified as compression, literary polishing, imagistic substitution, allusive rewriting, or factual distortion.

Revision needed: provide a compact methodological subsection explaining the comparison protocol. For instance:

• what counts as a narrative element,
• how omissions are identified,
• how exaggeration is distinguished from poetic condensation,
• and how the paper decides when an alteration has epistemic consequences rather than merely stylistic ones.

Without such clarification, evaluative claims like “hinders the reader’s direct understanding” or “distortion of the narrative’s factual accuracy” remain plausible but insufficiently standardized.

3. Control more carefully for genre as an alternative explanation.

The paper repeatedly argues that the shift from detailed record to generalized and embellished form reflects the influence of the “culture without soldiers.” Yet the manuscript itself also acknowledges a simpler and potentially powerful explanation: Bi’wo ji is a “narrative poem” and was therefore “compelled to adopt a highly generalized approach.” Similarly, one insertion is explained as having been added “purely to meet the metrical requirements of the poetic form.”

This creates a methodological problem. Some of the features attributed to historiographical culture may follow directly from genre constraints. If so, the paper must show why “culture without soldiers” explains more than poetic compression alone.

Revision needed: disentangle genre effects from historiographical culture. A useful way forward would be to categorize each example under one of three headings:

• likely caused by poetic form,
• likely caused by literati stylistic preference,
• likely caused by deeper historiographical inheritance.

At present, the article sometimes conflates these levels.

4. Moderate claims of factual distortion unless the baseline of factuality is defended.

A particularly strong claim appears in Section II: “The most serious consequence of such substitutions and insertions is the distortion of the narrative’s factual accuracy.” The example given is the shift from “the river’s flow grew shallower by the day” to “the river ran dry.” This is a good example, but the article assumes the Shoucheng lu as the stable baseline of factual truth. Elsewhere, however, the author acknowledges limits of verification, writing: “I am ultimately in no position to properly verify the merits, flaws, or accuracy of the powers of observation shown in the standard histories’ accounts of war.”

Revision needed: specify whether “distortion” means deviation from the earlier text, deviation from recoverable historical reality, or deviation from probable battlefield conditions. These are not equivalent. The current argument sometimes slides among them.

5. Strengthen engagement with counterexamples and comparative controls.

The paper argues for a developmental trajectory “from passive inheritance toward active re-creation.” This is intriguing, but the article would be more persuasive if it incorporated at least one control or counterexample:

• another pair of texts by the same author,
• another Song war text not rewritten into poetry,
• or an example where literary transformation preserved rather than reduced military intelligibility.

Without that, it is difficult to tell whether the Zhao Wannian case is typical, anomalous, or especially convenient.

6. Reconsider some broad civilizational or field-wide framing.

The manuscript cites Fairbank’s view that Chinese narratives lacked a “human and bellicose dimension” and Franke’s reduction of standard formulas. The author critiques these views to some extent, but the paper still risks reproducing overly sweeping binaries between “Chinese” and “Western” war writing. The invocation of Tolstoy, Stendhal, and Thackeray as contrast cases is suggestive but under-integrated into the argument.

Revision needed: either deepen the comparative framework or reduce it. As written, the comparisons are too brief to bear much interpretive weight and may distract from the much stronger textual analysis of the Chinese sources.

7. Tighten the article’s final epistemic position.

Near the end, the author writes: “If we do not fixate too narrowly on factual truth, we might instead do better to regard historians’ layered, accumulative writing as a form of re-creation rather than as a simple act of distorting text and history.” This is thoughtful, but it partially unsettles the article’s earlier criticism of distortion. The paper appears to oscillate between lamenting loss of factual precision and celebrating layered rewriting as historiographical vitality.

Revision needed: decide whether the article’s normative position is primarily:

• critical of literary mediation because it obscures war,
• or appreciative of re-creation as a productive historiographical mode.

A nuanced middle position is possible, but it must be stated more clearly.

Potential Minor Revisions

1. Stylistic and grammatical issues

The manuscript is generally polished, but several local issues should be corrected.

• In the Keegan quotation in the opening section, “military historians learnt to write their books from watching battles” appears in one place with the typo “fitht” in the note citation: “did not learn to fitht their battles.” This should be corrected.
• There is at least one spacing/run-together issue: “medieval historiansfrom the Han dynasty onward,” where a space is missing after “historians.”
• Another run-together issue appears in “tend toward seeking borrowings or even imitationsfrom pre-existing texts,” where a space is missing before “from.”
• A further typographic problem appears in “losyal sincerity pierced the sun,” where “losyal” should be “loyal.”
• There are visible OCR-like or encoding inconsistencies in several snippets, including broken hyphenation and character spacing. These may stem from file conversion, but they should be checked carefully in the submission version.

2. Formatting and presentation issues

• The article uses leaf numbers for primary-text citation, e.g., “leaf 21a,” “leaf 4b,” and “leaves 18b–19a.” This is workable, but the citation practice should be explained once clearly and consistently, especially for readers unfamiliar with such conventions.
• The manuscript includes substantial transliteration, Chinese characters, and translated titles. Ensure consistent styling for romanization, italics, and glosses.
• Some long paragraphs, especially in the introduction and reflection sections, should be broken into shorter units for readability.
• The title is interesting but somewhat broad relative to the actual case-study design. Consider a subtitle that foregrounds Zhao Wannian and the comparison of Xiangyang shoucheng lu and Bi’wo ji more explicitly.

3. Citation and source framing

• The article usefully identifies editions and textual witnesses, including the Zhihai edition and the Liang Song mingxian xiaoji version in Siku quanshu. This is a strength, but the source base would be easier to evaluate if the rationale for preferring certain witnesses were summarized more systematically.
• When the author notes that the Siku edition altered terms such as replacing lu with di, this is important and should be integrated more directly into the methodological implications of comparison.

4. AI content analysis for this post-2021 work

Estimated percentage of AI-generated content: 8–18%.

Basis for estimate:

• The manuscript shows sustained domain knowledge, coherent long-range argumentation, and specific philological control that are more characteristic of human scholarly composition than generic AI drafting.
• The prose is highly patterned and occasionally ornate, but not in the flattened, repetitive way commonly associated with obvious machine-generated academic text.
• There are a few signs that could reflect AI-assisted polishing or translation smoothing, such as abstract conceptual phrasing and elevated transitions, for example “developmental trajectory—from passive inheritance toward active re-creation” and “layered, accumulative writing as a form of re-creation.”
• No clearly machine-fabricated references or obviously hallucinated factual structures are apparent in the visible material.

Highlighted potentially AI-assisted sections:

• The abstract, especially its compressed conceptual sequencing and polished framing, reads like it may have undergone AI-supported language refinement, though not necessarily AI authorship.
• Some reflective prose in the conclusion, especially where the manuscript shifts from empirical case analysis to broad philosophical framing, also has a slightly synthetic register.

Assessed epistemic impact:

• Low to moderate. Even if AI-assisted polishing was used, the manuscript’s main contribution appears to rely on text-specific comparative scholarship rather than generic exposition. The larger risk is not fabrication but over-smoothing: AI-style polishing may intensify conceptual breadth without adding evidentiary discipline.

Recommendations

1. Reframe the article explicitly as a case study unless broader comparative evidence is added.
2. Insert a short methods paragraph defining how textual changes are identified and interpreted.
3. Separate genre constraint from historiographical culture throughout the analysis.
4. Use more cautious terminology than “distortion” unless the baseline of factuality is clearly justified.
5. Add at least one counterexample, control case, or brief comparative appendix/table of transformation types.
6. Clarify the article’s normative stance on rewriting: loss of military specificity, productive literary re-creation, or both.
7. Perform a final copyedit for spacing errors, typographic slips, and consistency in citation and transliteration.

Fraud / paper-mill screening note

I do not see strong paper-mill traits in the uploaded manuscript. The argument is coherent, the citations are field-appropriate, the source criticism is specific, and the examples are textually anchored. The main concern is not integrity but argumentative calibration: the manuscript’s strongest evidence is narrow and philological, while some claims are comparatively broad.

XSci (AI review)

September 26, 2026

Referee Report

Manuscript: "How Do Historians Learn to Write War? Shifts in Historical Understanding and Thought between the Record of the Defense of Xiangyang's City Wall and the Bi'wo ji"

Recommendation: Major revision

Summary and Recommendation

This article pairs two works by the Southern Song writer Zhao Wannian 趙萬年 — the day-by-day prose Xiangyang shoucheng lu 襄陽守城錄 (Zhihai text, leaves 1a–31b) and the verse Bi'wo ji 裨幄集 (Liang Song mingxian xiaoji, leaves 1a–5b) — and reads the differences between them as evidence for a "culture without soldiers" (wubing wenhua 無兵文化) operating in Chinese war writing. Having argued that the prose record came first, it tracks what is compressed, substituted and embellished as that record is reworked into commemorative verse, and proposes that the pattern exemplifies a literati tendency to understand war through pre-existing texts rather than through experience. It closes by recommending that such accretive rewriting be read as re-creation rather than corruption.

The comparative instrument is unusually well chosen and the philological work is in places acute. Matched accounts of the same events by the same author, separated by months rather than centuries and by register rather than information, are rare in the Song source base, and the author has done the unglamorous work of locating witnesses, establishing a relative chronology, and aligning passages leaf by leaf. What the draft does not yet establish is that the differences it documents are attributable to the cause assigned them. Three difficulties stand between this version and publication: a mismatch between the governing frame and the case chosen to demonstrate it; a conceptual apparatus that is asserted rather than defined and operates circularly; and an evidentiary procedure whose philological controls are assembled in Section I and then not brought to bear on the rest. None appears fatal, and several remedies lie within materials the author already has in hand. I recommend major revision.

Major Concern Category 1: The Fit between the Case Study and Its Governing Thesis

1.1 Self-revision is not the intertextual inheritance the frame describes

The opening frame, built from Keegan's remark about military historians who did not learn to write from watching battles, makes a specific claim: that because historians rarely experienced war firsthand they sought models in earlier accounts, which were then copied into later ones, producing a doubling or afterimage. Section IV restates this, and the Rogers comparison is likewise about Tang historians importing an account of one disaster into the record of another. The case study does not have that structure. Zhao Wannian was present at Xiangyang; the Shoucheng lu is his own record, the Bi'wo ji his own reworking of it. What Sections II and III document is self-revision across genres by an eyewitness, not a non-participant absorbing a predecessor's narrative. The afterimage metaphor depends on precisely that absence of access, and a writer with his own record before him has no such gap to fill. A related shift of unit passes unexamined: the analysis concerns one author revising himself, while Section IV generalizes to historians inheriting across generations.

1.2 Genre, metre, and the occasion of a victory song

Every phenomenon catalogued in Section II is a standard consequence of the form the second text takes. Compressing watchtowers, cannon and ox-hide shelters into "their own siege equipment," rounding three thousand four hundred to three thousand, substituting crossbows and forked pikes for enumerated troop categories, pairing goose-carts with tiger-cannons, and adding gongs to drums are what seven-character lines and parallel diction require. The article supplies this explanation twice itself, noting that "wild wind" was "inserted purely to meet the metrical requirements of the poetic form" and that riqian 日淺 "could not easily be reduced to a single character." Where those explanations hold, the cultural one does no additional work. No contrast case is specified — the article never says what a comparable rewriting by an author outside the culture without soldiers would look like — so any compression confirms the hypothesis. A further alternative goes unconsidered: the Que di kaige 卻敵凱歌 is a public, merit-claiming genre, and Section I records that Zhao was promoted to Wude dafu 武德大夫 for his service. Heroic idiom and distortions running uniformly in the defenders' favour are what a text claiming credit would produce.

1.3 The biographical premise that rescues the thesis is stipulated

Section III raises the obvious objection — that Zhao, unlike the armchair literati described, was a soldier — and answers with a distinction: he had personal experience of war but lacked experience of commanding one. Both halves are under-evidenced. The first rests on a single line of the Bi'wo ji itself, that the author rose through the military examinations and served in the ranks of war, which is a self-representation inside the very text argued to be embellished. For the second no evidence is offered. The article's own quotations pull against the premise: the wartime title in Section I, opening with mensheng 門生 and running through a concurrent vice-commandership and attachment to the Jingxi Beilu pacification office, describes a staff and client position close to the command rather than service in the ranks, and it is quoted for chronology alone. If Zhao wrote as a client of the commander he praises, patronage may explain the turn to panegyric as economically as a millennium of civil-over-martial disposition. The distinction should be grounded, the title analysed, the commander identified, and the claim stated falsifiably.

Major Concern Category 2: Conceptual Foundations and the Long-Durée Premise

2.1 "Culture without soldiers" is undefined and operates circularly

The concept carrying the explanatory burden is never stated. It enters as a bare reference to Lei Haizong's 1935 essay and is thereafter treated as a settled fact of Chinese civilization persisting for a millennium. The reader is not told what it asserts, at what level it operates, or what would count against it, and the article slides among readings, invoking it to explain a poet's diction, the marginalization of military affairs, and the behaviour of scholars as a class. Lei's essay also has a history worth acknowledging, written in an interwar idiom of national self-diagnosis by a figure later central to the Zhanguo ce circle; a concept forged as a diagnosis of national weakness is converted here into a neutral descriptive category without argument. Most seriously, the concept operates in a closed circle: the compression observed is explained by the culture without soldiers, while that culture is evidenced by nothing other than such compression. Three rivals — generic constraint, commemorative occasion, and the ordinary economy of any second telling — would predict the same outcomes and are never entertained.

2.2 The Chunqiu and Gongyang genealogy will not bear the weight placed on it

The long-durée premise rests on jin zhanfa er wu wuci 謹戰伐而惡無辭, drawn from the Dong Zhongshu exchange in the Chunqiu fanlu and defended by Duan Xizhong. Note 4 candidly records that jin zhanfa appears to have been coined by Duan himself, yet Section II attributes the formulation directly to the Chunqiu fanlu, so that a modern coinage systematizing a Han exegetical claim about the ordering of names becomes a description of practice across a millennium of standard histories. The illustration compounds this: none of the three Yin 1 entries quoted is an interstate campaign, the third being an internal fratricidal conflict. That entry is moreover the locus classicus of praise-and-blame reading, since the commentarial traditions take ke 克 to carry censure — rendering it "defeated" rather than "overcame" suppresses the evaluative charge that makes it a counter-instance. A Gongyang hermeneutic, a formal property of annalistic entries, and the habits of biographical historiography are treated as one, and the expansive Zuozhuan, which Section III itself calls a chopping-and-hacking book, is absorbed rather than explained. The excellent criticism that Fairbank and Franke are skewed by relying solely on annalistic form also sits awkwardly beside Section II's adoption of that same baseline.

2.3 The developmental trajectory is asserted rather than demonstrated

The abstract promises that close analysis identifies a trajectory from passive inheritance toward active recreation, and Section IV restates it. Yet nothing in Sections II or III is diachronic: the evidence is paired passages from two works by one author on the same events, composed within roughly a year of each other. No earlier stage of passive inheritance is exhibited, no later stage of active recreation, and no mechanism is specified by which one becomes the other. The materials that might have supplied a trajectory — Rogers on the Fei River, Zhou Yiliang on Wei Shou and on the Southern and Northern Dynasties, and the responses of Hu Baoguo and Qian Maowei — arrive only after the analysis has concluded and are summarized rather than put to work. The boldest thesis therefore rests on the section containing no new evidence, while the section containing the evidence supports a much more local conclusion. Either the trajectory should be scaled back to a labelled hypothesis, with the abstract revised to match, or further cases at different points in time would be needed.

2.4 The comparative apparatus is illustrative, and four claims need correction

The cross-cultural reach is among the article's most distinctive features, which is why the thinness of its execution matters. Clausewitz, then serving with the Russian army, held no command at Borodino that Tolstoy could have diminished; his one appearance in War and Peace is a brief satirical scene, and Berlin's essay concerns Tolstoy's philosophy of history rather than Clausewitz. The stronger point evidently wanted — that Tolstoy displaces the commander's-eye view in favour of the participant's — is directly germane and is what the cited works would support. Vanity Fair, second, famously declines to narrate Waterloo, disposing of George Osborne in a sentence and keeping the reader among civilians in Brussels; that concerns the home front rather than soldiers' experience, and Stendhal's Fabrice is the apposite case. The Yongbi eocheonga, third, was compiled under Sejong to celebrate six royal ancestors with the founder Taejo foremost, though the canto cited does fall in the Taejong material. Fourth, the closing clause describes Keegan as lame from birth; he contracted orthopaedic tuberculosis at thirteen, so the disability was acquired rather than congenital. As that clause carries the rhetorical weight of the recuperative argument, the correction matters more than its length suggests.

Major Concern Category 3: Evidentiary Procedure and Philological Control

3.1 The relative dating is circular, and the transmission history is not brought to bear

Section I rightly treats the relative chronology as decisive, and candidly reports that the external notices order the two titles differently and say only that the works were written "in the army." Of the three arguments offered, the first — that the prose record is colloquial whereas the verse is embellished — assumes the direction of development the article sets out to prove. The second, as printed, states that the Shoucheng lu shows clear signs of having been polished in comparison with the Bi'wo ji, which reverses the thesis and appears to be a slip; as written it is also self-cancelling, since the first argument rests on the prose record being unpolished. The third is independent and good, though an office title in the opening matter dates the opening matter rather than the whole text, and the genealogy supporting the promotion, reproduced in a 2004 reprint, is of unassessed reliability. A parallel problem attends the variants themselves. Neither text survives in a Song witness; the Bi'wo ji comes through Siku anthologies already twice mediated, and note 14 concedes possible Qing supplementation while Section I documents a systematic replacement of lu 虜 by di 敵. Once editorial hands between the thirteenth century and the nineteenth are on the record, the step from a lexical difference to an authorial choice needs an argument — most pressingly for the smallest and most load-bearing examples, pingming 平明 for weixiao 未曉, congzi 從茲 for zici 自此, the added zheng 鉦, and the inserted willow shade.

3.2 The announced collation is not reported, and the collection is not described

Section I promises philological control the rest of the article does not draw upon: a Nanjing University copy preserving the original to a certain degree, a uniform restoration of lu for di, cross-checking against the Qing manuscript in Song ji zhenben congkan, and an Academia Sinica manuscript longer than either Siku witness. No variant reading from any of these appears in Sections II to IV, and note 14 states that the article continues to cite the Siku edition. Readers are thus asked to accept roughly fifteen comparisons on the authority of a single witness the article has itself characterized as altered. A short collation table, one row per passage and one column per witness, would suffice; the grounds for setting aside the Academia Sinica manuscript also conflate the irrelevance of the appended Zhu Lü shimo ji with the relevance of that manuscript's readings for the Bi'wo ji. A second gap compounds this: the introduction calls the Bi'wo ji a narrative poem while Section I treats it as a collection, and quotations from leaf 1b, leaf 3a and leaves 3b–5b are pooled without distinction across at least two compositions. Since the article argues that material was inserted purely to fill out a seven-character line, the metre of each quoted line is essential and is nowhere given.

3.3 Borrowing is not distinguished from availability, and the sample is undocumented

Section III carries the central claim and rests on demonstrations that turn on ubiquity: that yin feng zonghuo 因風縱火 recurs in the Shiliuguo chunqiu, Shuijing zhu, Song shu and Chen shu, and that xing wu guang 星無光 recurs throughout Tang and Song astronomical writing. But ubiquity disqualifies a phrase as the trace of one author reading one text; a formula distributed across four works is a fact about the diction available to any literate Southern Song writer, as the article half concedes in noting that the second phrase was no less frequent than the first. Absent are evidence that Zhao read any of the named works, a frequency baseline, a control set of phrases not traceable to earlier texts, and a threshold at which recurrence becomes allusion — all feasible with searchable corpora. The willow instance is weakest, resting on a single line by Yan Shu for one of the commonest images in the Tang–Song lexicon, with a topographical explanation never excluded. The sample is likewise undocumented, and the article's own counter-instances are narrated rather than counted: the xige reading, in which the later text supplies information the earlier lacks, and the addition of "deep winter," which given a winter siege may be accurate supplementary fact inside a passage read as a case of loss.

3.4 The prose record is an uninterrogated benchmark, and no method is stated

Every divergence is measured against the Shoucheng lu in a normative vocabulary of ambiguity, sacrificed accuracy and distortion, each requiring that the prose record give unmediated access to what happened. Section I supplies reasons to doubt this: the record was begun during the war and completed shortly after, by a man whose merit was being assessed and who was promoted for it. Such an account is shaped and interested in the same way the verse is, and its quantities are of the kind such documents inflate — the several tens of thousands drowned at leaf 5b is accepted as the baseline against which a later addition registers as embellishment. If both texts are self-representations, the differences record a shift of rhetorical register rather than a decline from fact into literature. The article half knows this, since its closing pages propose reading rewriting as re-creation while Sections II and III are written in the opposite key, so that Section IV reads as a partial retraction. Compounding this, no paragraph anywhere states the corpus compared, the procedure by which parallels were identified, the criteria separating borrowing from common diction, or what would have counted against the thesis; Sections II and III do not observe the distinction their titles promise, and Section IV simultaneously generalizes, introduces the material carrying the largest claims, and concludes. There is no historical-context section and no separate conclusion.

Minor Issues

1. Section I states that the Que di kaige at leaves 3b–5b occupies more than half of the collection's length. Since the collection runs 1a–5b, that is five half-leaves of ten, approximately half. If front matter occupies part of the opening leaves, saying so would resolve the discrepancy; a sentence on leaves 2a–2b, from which nothing is quoted, would also help.

2. No dates appear anywhere in the body — not Zhao Wannian's, not the siege's, not either work's — the only dates being embedded in the title of Avery's thesis. This is striking given that a whole section concerns relative chronology, and it compounds the absence of orientation on the Kaixi war, Han Tuozhou's campaign, and the defending commander, who appears only as "the lord."

3. The texts' relative extents, thirty-one leaves against five, appear only in the notes, although the argument about loss of detail depends on that ratio. Chinese text is also supplied only intermittently, so that where an argument turns on a single graph — wei 猬, he 涸, zheng 鉦, riqian 日淺 — readers cannot check the claim against the original.

4. The verdict that Zhao "had no choice but to exaggerate the fact" to a river gone dry depends on taking he 涸 in its strongest sense. In classical prose he is used of water running low as well as of complete desiccation; on the weaker reading the two accounts are nearly compatible, and the strongest claimed instance of factual distortion largely dissolves.

5. Leaf 18a gives three thousand four hundred troops dispatched, while leaves 18b–19a enumerate more than fifteen hundred men with digging tools and more than six thousand six hundred crossbowmen, plus militia and men with forked sickles and axes. Whether these describe the same operation is not stated; if they do, the totals require reconciliation.

6. The ambiguity attributed to weiji 猬集 is not well supported by the line as quoted, which runs "Arrows massed like hedgehogs, their barbed points bristling." The word "arrows" names the referent and "barbed points bristling" restores the bristle image mao carried, so the couplet appears less ambiguous than the claim requires.

7. The omission of "Guanghua" is offered as evidence of abandoning place names, but the comparison sets a three-clause prose sentence against a seven-character line retaining two of three names. Dropping one clause of three is what compression looks like; demonstrating tendency would require an instance where a name is dropped although space permitted it.

8. Bi'wo ji quotations carry leaf numbers but no line positions, which weakens the parallelism arguments: the claim that "cannon" was extended into "tiger-cannon" to echo "goose-carts" cannot be assessed without the couplet. Section II's statement that the verse was "compelled to adopt a highly generalized approach" because the prose record is detailed also leaves the causal direction unclear.

9. Several notes would repay attention. Note 7 reverses the editors of Chinese Ways in Warfare, conventionally Frank A. Kierman, Jr. and John K. Fairbank. Note 15 gives Chen Naigan's Dushu shixiao lu without place, publisher, date or serial, though it underwrites a genuine bibliographical finding. Note 29 cites McMullen across twenty pages plus notes, and "this work" there most naturally reads as McMullen although Tao's 1970 review is of Rogers. Note 30 gives Twitchett's article without year or page range. Note 13's juan numbering would benefit from reconciliation with the 380-juan recension Xu Hongxia studies, and no dates are given for either collectaneum, so the claim that the Zhihai is the earlier edition cannot be checked.

10. Presentation could be regularized. The abstract's promise to reconsider the role of war narrative in the Chinese historiographical tradition exceeds what the analysis delivers, and the keywords name none of the article's actual objects; the Keegan epigraph is reproduced in full three times. Romanization varies (粵雅堂叢書 as "Yueyang congshu" rather than Yueyatang congshu), and several graphs appear corrupted, among them 錢熊祚 for 錢熙祚, 段熊仲 for 段熙仲, 晟殊 for 晏殊, 余嘉錡 for 余嘉錫, and 裨幣 for 裨幄; "losyal sincerity" is a slip. Li Xinchuan, Yuan Hong, Shen Yue, Wei Shou and Yan Shu are introduced without identification, where Zhang Xun and Xu Yuan receive a gloss. It would also help to state on what authority the lu/di restorations were made, and whether the Zhihai editors abridged their exemplar.

Missing and Inadequately Integrated Literature

Cited but Inadequately Integrated

Lei Haizong's "Wubing de wenhua" (1935) supplies the article's load-bearing concept and appears once, in a footnote, without discussion or definition; it needs a paragraph in the body. Nicola Di Cosmo's introduction to Military Culture in Imperial China is cited twice for two short phrases, one of which then does heavy lifting across the abstract and three sections, although the volume was assembled precisely to complicate the Fairbank picture the article accepts, and its disaggregation of military culture into distinct levels is the level-specification the argument needs. David A. Graff's "Narrative Maneuvers," in the same volume, is cited once for two stock phrases although it is the closest existing model for this method and should be the principal interlocutor.

Julie J. Avery's translation of the Shoucheng lu is engaged at exactly one point, to dispute the rendering of niuwu, where a complete translation is the natural comparison for every prose passage discussed. Yu Jiaxi's Siku tiyao bianzheng entry is cited bibliographically but never for its philological findings, though Section I is an edition-history argument, and the same applies to Xu Hongxia, whose study of the transmission of the Xiaoji appears in a concessive clause when it should govern the article's procedure. Li Xinchuan's assessment of accounts of the Battle of Caishi by association with known literary exaggerations is a close analogue to the article's own method and deserves a paragraph, since it supplies a native criterion for when formulaic language signals exaggeration. Jing-Shen Tao's review and David McMullen's book are cited to establish that criticism of Rogers exists without summarizing it, and Valerie Hansen is cited for three pages to supply the spectrum of source types on which the whole design depends.

Truly Missing Papers

Yuan-kang Wang (2011) challenges the cultural-pacifism reading on substantially Song evidence, arguing that Confucian pacifism was the prevailing strategic culture but did not govern strategic behaviour — a counterweight to the inference from literary disposition to civilizational fact. Peter Lorge (2005, 2015) argues that Song civil dominance was a product of tenth-century warfare rather than a founding principle, bearing on the assertion that opens Section II. Charles Hartman (2021) is now standard for Song source criticism, and his account of how surviving narratives reflect the circumstances of their compilation is close kin to the notion of textual doubling. Susan Cherniack (1994) supplies a framework for distinguishing authorial from editorial variation. David Schaberg (2001) and Wai-yee Li (2007) would discipline the handling of the Chunqiu and Zuozhuan, particularly the assumption that terseness signals absence of evaluation. Christopher M. B. Nugent (2010) establishes how much of a poem's diction may derive from shared stock without any specific act of borrowing. Paul Fussell (1975) and Samuel Hynes (1997) are the closest Western analogues to the thesis, Hynes especially, since he studies the participant who was present and who then reaches for available literary forms. Richard L. Davis (2009) would anchor the historical-context section the article lacks.

Cherniack, S. 1994. Book Culture and Textual Transmission in Sung China. Harvard Journal of Asiatic Studies 54:5–125.

Davis, R. L. 2009. The Reigns of Kuang-tsung and Ning-tsung. In The Cambridge History of China, vol. 5, pt. 1. Cambridge University Press.

Fussell, P. 1975. The Great War and Modern Memory. Oxford University Press.

Hartman, C. 2021. The Making of Song Dynasty History: Sources and Narratives, 960–1279 CE. Cambridge University Press.

Hynes, S. 1997. The Soldiers' Tale: Bearing Witness to Modern War. Allen Lane.

Li, W. 2007. The Readability of the Past in Early Chinese Historiography. Harvard University Asia Center.

Lorge, P. 2005. War, Politics and Society in Early Modern China, 900–1795. Routledge.

Lorge, P. 2015. The Reunification of China: Peace through War under the Song Dynasty. Cambridge University Press.

Nugent, C. M. B. 2010. Manifest in Words, Written on Paper. Harvard University Asia Center.

Schaberg, D. 2001. A Patterned Past: Form and Thought in Early Chinese Historiography. Harvard University Asia Center.

Wang, Y. 2011. Harmony and War: Confucian Culture and Chinese Power Politics. Columbia University Press.

Conclusion and Path Forward

This is a project worth completing. The pairing of the two texts is a clever piece of source work on material that deserves more attention than it has received; the edition history in Section I reflects real archival labour, and the identification of the Academia Sinica manuscript with the transcription Yu Jiaxi cited only at second hand is a bibliographical contribution meriting more prominence than a single clause. The reading of niuwu through xige is a genuine small discovery that corrects a published translation, and the observation that inserting yin feng zonghuo converts a wind that happened to rise into a wind deliberately exploited is an excellent catch that survives even if the borrowing claim around it is withdrawn. Most promising of all is the closing proposal that layered writing be read as re-creation rather than distortion — the most original idea here, and the one the evidence best supports.

The most productive revision would narrow rather than broaden. Presented as a study of what happens to eyewitness testimony when it migrates from prose record into commemorative verse, with implications for using Song poetry as historical evidence, and with the wubing wenhua material as context rather than conclusion, this would be a publishable and useful article. Concretely, that means correcting the second dating argument and acknowledging the chronology's probabilistic character; defining the governing concept and stating what would disconfirm it; supplying a statement of method, a historical-context section, and an appendix tabulating every alignable parallel; reporting the collation already carried out; acknowledging that the prose record is itself an interested document and revising the vocabulary of loss accordingly; and either building the comparative frame into a real comparison or reducing it, with the four factual corrections made. These are substantial but tractable revisions, and none requires abandoning the article's real insight.