The user wants me to convert an OCR image of an engineering standard PDF page into an HTML fragment. Let me analyze the page carefully.
This is page 222R-2 from an ACI Provisional Standard. It's a two-column layout.
- A figure (Fig. A1—Soxhlet extraction apparatus) with labels: CONDENSER, GLASS WOOL, SAMPLE HOLDER, WATER LEVEL, SAMPLE, SYPHON TUBE
- Caption: "Fig. A1—Soxhlet extraction apparatus"
- Continuation of 2.2: "measured is very dependent on the degree of fineness to which the aggregates are ground during sample preparation.† The problem with the ASTM C 1218 test method is therefore twofold: the test measures chlorides that are not generally available for the corrosion reaction, and the test gives widely variable results. The test method described herein should be used when chloride-bearing aggregates influence the results obtained using ASTM C 1218."
- 2.2—Sulfides are known to interfere with the determination of chloride content. Blast-furnace slag aggregates and cement contain sulfide sulfur in concentrations high enough to cause significant interference and produce erroneous test results. Treatment with hydrogen peroxide, as discussed in ASTM C 114,‡ shall be used to eliminate such interference.
- 3.1—Sampling equipment: ...
- *ASTM Standard Test Method C 1218-92, Standard Test Method for Water-Soluble Chloride in Mortar and Concrete.
- †For more information see "The Determination of the Chloride Content of Concrete" by Brian B. Hope, John A. Page, and John S. Poland, Cement and Concrete Research, V. 15, No. 5, Pergamon Press, New York, Sept. 1985, pp. 863-870.
- ‡ASTM Test Method C 114, "Test Methods for Chemical Analysis of Hydraulic Cement."
- §ASTM C 42, "Test Method for Obtaining and Testing Drilled Cores and Sawn Beams of Concrete."
- **Suitable Soxhlet extraction equipment is available from Fisher Scientific (Catalog No. 09-551A) and other manufacturers.
Now, the page header "222R-2" and "ACI PROVISIONAL STANDARD" should be ignored (filter rule 1).
The figure itself should be ignored, only the caption text.
The page number at bottom - there's no visible page number at the bottom actually. The footnotes are at the bottom. The page number "222R-2" is at the top (header), so it's ignored. But the output must end with
Now let me think about the two-column layout. The instructions say two-column layout →
. But this is a continuous text flow across columns. Hmm. The text flows from left column to right column. Actually the figure is in the left column top, then text continues below it. The right column has text that continues from left column's 2.2.
This is tricky. The two-column layout with continuous flow. I think the best approach is to use the layout-columns table to represent the two columns, but the text flows continuously. Actually, since the text flows from left to right column (2.2 continues), putting them in separate table cells would break the flow.
Let me think. The instruction says "两栏布局 →
". But for a flowing two-column document, it's hard. However, the figure is in the left column. The text in left column: figure caption, then 1 Scope, 2 Significance and Use (2.1, 2.2 start). Right column: 2.2 continuation, 2.2 sulfides, 3 Apparatus, 4 Reagents, 5 Sampling.
Hmm, but the 2.2 paragraph spans both columns. If I put left column content in one td and right in another, the 2.2 paragraph would be split awkwardly.
Actually, maybe the cleanest faithful representation: Since this is a flowing two-column layout, perhaps I should NOT use the layout-columns table for the body text, but just present it as a single flow? But the figure is positioned in the left column.
Let me reconsider. The rules say "两栏布局 → table class layout-columns". This is a排版替代方案 for when there's a genuine two-column layout. But for a standard flowing document, the reading order is left column top-to-bottom then right column top-to-bottom. The figure is at the top of the left column.
I think the most faithful approach given the constraints: Use the layout-columns table to hold the figure (left cell) and... no, that doesn't work because the right column has lots of text.
Hmm, let me think about what's most reasonable. The figure is a block element in the left column. The text flows. I think I'll present the content in reading order as a single flow (since it's a continuous article), and place the figure caption where it appears. But the figure is at the top of the left column, before "1 Scope". In reading order, the figure caption comes first, then 1 Scope, etc. That works fine as a single flow!
Wait, but then the right column text (2.2 continuation) comes after the left column's 2.2 start. In reading order that's exactly correct: left column fully, then right column fully. So a single linear flow preserves reading order perfectly. The only thing is the visual two-column layout is lost, but the content order is preserved.
But the instruction explicitly provides a two-column layout alternative. Should I use it? The problem is the text flows across columns, so a table with two cells would break a paragraph across cells, which is bad.
Let me look at the structure again. Left column: figure + caption, then 1 Scope (1.1, 1.2), then 2 Significance and Use (2.1, 2.2 first part). Right column: 2.2 second part, 2.2 sulfides, 3, 4, 5.
If I use layout-columns table: left td = figure caption + 1 + 2.1 + 2.2(part1), right td = 2.2(part2) + 2.2 sulfides + 3 + 4 + 5. The 2.2 paragraph would be split into two cells. That's not ideal but it's how the visual layout is. However, splitting a paragraph across table cells is semantically wrong and the rules say "标签必须正确闭合，嵌套必须正确" and "不要重复同一段文字". Splitting a paragraph means I'd have two
tags for one paragraph, which is acceptable as long as I don't duplicate text. But it's awkward.
Hmm. Actually, I think for OCR of a flowing two-column page, the standard practice is to linearize. The two-column layout alternative is meant for things like side-by-side lists or a figure next to text where they're independent. Here the columns are a continuous flow.
But wait — the figure is independent (it's a figure). The text flows. So maybe: the figure caption is a standalone block. Then the text is a continuous flow.
Let me look at the example given in the prompt. The example doesn't use layout-columns at all. It just has h2, p, table. So linear flow is acceptable.
I'll go with linear flow in reading order. That's the safest for content fidelity and avoids splitting paragraphs.
Actually, let me reconsider whether to use layout-columns. The visual layout is two columns. The instruction lists it as an option. But using it would force splitting the 2.2 paragraph. I think linear flow is better and the example supports it. I'll do linear flow.
Hmm, but actually, maybe I should use layout-columns to capture the visual structure, putting the figure in left and... no. Let me just do linear. The reading order is what matters for content.
Wait, but there's a subtlety: in the left column, the figure occupies the top portion, and the text "1 Scope" starts below the figure. In the right column, text starts at the very top. So visually, the right column's top text (2.2 continuation) is at the same vertical level as the figure. But in reading order, the figure caption and left text come first. That's fine.
OK here's my decision: linear flow. Figure caption first (since it's at top of left column), then 1 Scope, 2 Significance, etc.
Hmm, but actually the figure caption "Fig. A1—Soxhlet extraction apparatus" — should it come before "1 Scope"? In the left column, the figure is at top, caption below it, then "1 Scope" below that. Yes, caption first. Good.
Now let me also handle the footnote markers. The text has †, ‡, §, *, ** markers. I should keep them. The rules say keep footnotes as small-note at original position. The footnotes are at the bottom of the page. I'll output them as small-note paragraphs at the end (before page number). The markers in the body text: I'll keep them as plain characters (†, ‡, §, *, **). Could use sup? The markers appear as superscript in the original († after "preparation", ‡ after "C 114", § after "C 42", * after "system", ** after "available"). I could wrap them in . The allowed tags include sup. That would be faithful. Let me use sup for the footnote markers in the body. Actually the asterisk * in "system.*" and ** in "available.**" are superscript. And †, ‡, § are superscript. Yes, use sup.
Let me check the body text for markers:
- "sample preparation.†" → sup
- "ASTM C 114,‡" → sup
- "ASTM C 42.§" → sup
- "cement system.*" → sup
- "equipment is available.**" → sup
Good.
Now the footnotes at bottom:
*ASTM Standard Test Method C 1218-92, Standard Test Method for Water-Soluble Chloride in Mortar and Concrete.
†For more information see "The Determination of the Chloride Content of Concrete" by Brian B. Hope, John A. Page, and John S. Poland, Cement and Concrete Research, V. 15, No. 5, Pergamon Press, New York, Sept. 1985, pp. 863-870.
‡ASTM Test Method C 114, "Test Methods for Chemical Analysis of Hydraulic Cement."
§ASTM C 42, "Test Method for Obtaining and Testing Drilled Cores and Sawn Beams of Concrete."
**Suitable Soxhlet extraction equipment is available from Fisher Scientific (Catalog No. 09-551A) and other manufacturers.
These are footnotes. Output as small-note. The markers at start of each footnote: I'll keep them as plain text at start (e.g., "*ASTM Standard..."). Actually in the original the marker is superscript at the start. I'll just put the marker as plain text. Or sup. The footnote text starts with the marker. I'll keep marker as plain leading character to be safe, or sup. Let me keep as plain text leading char since it's the footnote label. Hmm, but to be faithful, the marker is superscript. I'll use sup for the leading marker too. Actually simpler: just plain text. The rules don't forbid plain text. I'll keep markers as plain text in footnotes for clarity. Hmm, but in body I used sup. For consistency maybe sup in footnotes too. Let me use sup for the leading marker in footnotes as well. Actually, let me keep it simple and use the marker as plain text at the start of each small-note. That's fine and readable.
Wait, the rule: "脚注 → 保留，按原文位置输出为
...
". The original position of footnotes is at the bottom of the page. So I place them at the end. Good.
Now the figure: ignore image, output caption only. Caption: "Fig. A1—Soxhlet extraction apparatus". It's italic in original. I'll wrap in em. The caption is left-aligned (it's at left margin under figure). I'll just use a p with em. Should it be align-center? In the image the caption "Fig. A1—Soxhlet extraction apparatus" is left-aligned under the figure (the figure is centered in left column but caption is left-aligned). I'll keep it as a normal p with em. Actually it looks left-aligned. Fine.
Now headings:
- "1—Scope" → h2 (page 1 of this fragment? The rule says page内标题 first page h2. This is the first page of the fragment. So h2 for top-level sections.)
- "2—Significance and Use" → h2
- "3—Apparatus" → h2
- "4—Reagents" → h2
- "5—Sampling" → h2
- Subsections 1.1, 1.2, 2.1, 2.2, 3.1, 3.2, 3.2.1, 3.2.2, 4.1, 5.1 are paragraphs with bold/numbered lead-ins, not headings. In the original they're run-in bold text at start of paragraph. I'll render as
1.1—...
. The numbers like "1.1—" are bold in original. Yes bold.
Let me check bold: "1—Scope" heading bold. "1.1—This test method..." the "1.1—" is bold. "1.2—This test method does not purport..." the "1.2—" bold and the rest italic. "2—Significance and Use" bold heading. "2.1—Water-soluble chloride..." bold "2.1—". "2.2—Sulfides..." bold "2.2—". "3—Apparatus" heading. "3.1—Sampling equipment:" bold "3.1—" and "Sampling equipment:" italic. "3.2—Sample processing apparatus:" bold "3.2—" and italic rest. "3.2.1 Samples too large..." bold "3.2.1". "3.2.2 Extract chlorides..." bold "3.2.2". "4—Reagents" heading. "4.1—The reagents..." bold "4.1—". "5—Sampling" heading. "5.1—Reduce the size..." bold "5.1—".
Good.
Now the 2.2 paragraph that spans columns: In linear flow, left column has 2.1 then 2.2 first part: "2.2—Water-soluble chloride, when present in sufficient amounts, may initiate or accelerate the corrosion of metals such as steel embedded in or contacting a cement system such as mortar, grout, or concrete. Other test methods exist for the determination of water-soluble chloride in a cement system.* However, some aggregates contain a considerable amount of chloride that is bound in the aggregate and is not available for the corrosion reaction. The test method described in ASTM C 1218 measures a portion of the chloride contained in these aggregates. However, the amount of chloride" then right column continues "measured is very dependent on the degree of fineness to which the aggregates are ground during sample preparation.† ..."
So in linear flow, I need to merge these into one paragraph. The left column's 2.2 ends with "the amount of chloride" and right column begins "measured is very dependent...". So combined: "...the amount of chloride measured is very dependent on the degree of fineness to which the aggregates are ground during sample preparation.† The problem with the ASTM C 1218 test method is therefore twofold: the test measures chlorides that are not generally available for the corrosion reaction, and the test gives widely variable results. The test method described herein should be used when chloride-bearing aggregates influence the results obtained using ASTM C 1218."
Yes! That's one continuous paragraph. Good, linear flow handles it perfectly.
Then a separate paragraph "2.2—Sulfides are known to interfere..." Wait, that's also numbered 2.2? Let me look. In the image, after the first 2.2 paragraph (which ends "...ASTM C 1218."), there's a new paragraph starting "2.2—Sulfides are known to interfere with the determination of chloride content." Hmm, that's odd, two 2.2 paragraphs? Let me look carefully.
Looking at the image: Left column bottom: "2.2—Water-soluble chloride, when present in sufficient amounts, may initiate or accelerate the corrosion of metals such as steel embedded in or contacting a cement system such as mortar, grout, or concrete. Other test methods exist for the determination of water-soluble chloride in a cement system.* However, some aggregates contain a considerable amount of chloride that is bound in the aggregate and is not available for the corrosion reaction. The test method described in ASTM C 1218 measures a portion of the chloride contained in these aggregates. However, the amount of chloride"
Right column top: "measured is very dependent on the degree of fineness to which the aggregates are ground during sample preparation.† The problem with the ASTM C 1218 test method is therefore twofold: the test measures chlorides that are not generally available for the corrosion reaction, and the test gives widely variable results. The test method described herein should be used when chloride-bearing aggregates influence the results obtained using ASTM C 1218."
Then new paragraph: "2.2—Sulfides are known to interfere with the determination of chloride content. Blast-furnace slag aggregates and cement contain sulfide sulfur in concentrations high enough to cause significant interference and produce erroneous test results. Treatment with hydrogen peroxide, as discussed in ASTM C 114,‡ shall be used to eliminate such interference."
Hmm so there are two paragraphs both labeled 2.2? That seems like an OCR/standard quirk. Wait, maybe the first one is 2.1? Let me re-examine. Left column: "2—Significance and Use" heading. Then "2.1—Water-soluble chloride, when present in sufficient amounts, may initiate or accelerate the corrosion of metals such as steel embedded in or contacting a cement system such as mortar, grout, or concrete. Other test methods exist for the determination of water-soluble chloride in a cement system.* However, some aggregates contain a considerable amount of chloride that is bound in the aggregate and is not available for the corrosion reaction. The test method described in ASTM C 1218 measures a portion of the chloride contained in these aggregates. However, the amount of chloride"
Wait! Let me look at the image text more carefully. The left column after "2—Significance and Use" heading:
"2.1—Water-soluble chloride, when present in sufficient amounts, may initiate or accelerate the corrosion of metals such as steel embedded in or contacting a cement system such as mortar, grout, or concrete. Other test methods exist for the determination of water-soluble chloride in a cement system.* However, some aggregates contain a considerable amount of chloride that is bound in the aggregate and is not available for the corrosion reaction. The test method described in ASTM C 1218 measures a portion of the chloride contained in these aggregates. However, the amount of chloride"
Hmm, is it 2.1 or 2.2? The image shows "2.1—Water-soluble chloride..." Let me look. The text in the image left column: I see "2.1—Water-soluble chloride, when present in sufficient amounts..." Actually I need to look at the rendered text. The image shows after "2—Significance and Use":
Line: "2.1—Water-soluble chloride, when present in sufficient"
Yes I believe it's 2.1. Let me look at the OCR text in the image. The image text reads:
"2—Significance and Use
2.1—Water-soluble chloride, when present in sufficient amounts, may initiate or accelerate the corrosion of metals such as steel embedded in or contacting a cement system such as mortar, grout, or concrete. Other test methods exist for the determination of water-soluble chloride in a cement system.* However, some aggregates contain a considerable amount of chloride that is bound in the aggregate and is not available for the corrosion reaction. The test method described in ASTM C 1218 measures a portion of the chloride contained in these aggregates. However, the amount of chloride"
Yes, it's 2.1. Good. Then right column continuation of 2.1, then "2.2—Sulfides are known to interfere...". That makes sense! 2.1 is about water-soluble chloride and ASTM C 1218, 2.2 is about sulfides.
So the first big paragraph (spanning columns) is 2.1. Good.
Let me re-read the right column top: "measured is very dependent on the degree of fineness to which the aggregates are ground during sample preparation.† The problem with the ASTM C 1218 test method is therefore twofold: the test measures chlorides that are not generally available for the corrosion reaction, and the test gives widely variable results. The test method described herein should be used when chloride-bearing aggregates influence the results obtained using ASTM C 1218."
Yes that's the continuation of 2.1. Good.
So 2.1 paragraph = left part + right part merged.
Then 2.2 paragraph = sulfides.
Good.
Now let me also double check 1.2 italic: "1.2—This test method does not purport to address all of the safety problems, if any, associated with its use. It is the responsibility of the user of this test method to establish appropriate safety and health practices and determine the applicability of regulatory limitations prior to use." The whole 1.2 paragraph is italic. So
1.2—This test method does not purport to address all of the safety problems, if any, associated with its use. It is the responsibility of the user of this test method to establish appropriate safety and health practices and determine the applicability of regulatory limitations prior to use.
.
Now 3.1: "3.1—Sampling equipment: The apparatus required for obtaining samples by coring or sawing is described in ASTM C 42.§ Sampling by drilling is not applicable for this test and shall not be used." The "Sampling equipment:" is italic. So
3.1—Sampling equipment: The apparatus required for obtaining samples by coring or sawing is described in ASTM C 42.§ Sampling by drilling is not applicable for this test and shall not be used.
3.2: "3.2—Sample processing apparatus:" italic.
3.2—Sample processing apparatus:
— it's a run-in heading with no following text on same line? In the image, "3.2—Sample processing apparatus:" is on its own line (italic), then 3.2.1 follows. So it's like a subheading. I'll make it a p with strong+em. Could also be h3? The numbering 3.2 is a subsection. But visually it's bold-italic run-in. The rule says page内标题 h2 or h3 based on visual hierarchy. 3.2 looks like a subheading (italic, on its own line). Hmm. But 3.1 is also a run-in with text following. 3.2 has no text following on same line. I think treating 3.2 as a paragraph with bold-italic is fine, consistent with 3.1. I'll keep as p.
Actually, to preserve hierarchy, maybe 3.2 should be h3? But 3.1 is not a heading (it has body text). Inconsistent. I'll keep all numbered subsections as paragraphs with bold lead-in. That's the standard ACI style (run-in bold). Good.
Now 3.2.2 mentions "shown in Fig. A1." Keep as text.
Now let's write the figure caption. The figure labels (CONDENSER, GLASS WOOL, etc.) are part of the image, ignore them. Only caption.
Now the page number at bottom: there's no page number visible at bottom (the bottom has footnotes). The header "222R-2" is page header, ignored. The output must end with
2
. Good.
Now let's assemble.
Order:
1. Figure caption (em)
2. h2 1—Scope
3. p 1.1
4. p 1.2 (italic)
5. h2 2—Significance and Use
6. p 2.1 (merged)
7. p 2.2 (sulfides)
8. h2 3—Apparatus
9. p 3.1
10. p 3.2 (italic)
11. p 3.2.1
12. p 3.2.2
13. h2 4—Reagents
14. p 4.1
15. h2 5—Sampling
16. p 5.1
17. footnotes as small-note
18. page number
Wait, but the figure caption placement: In reading order, the figure is at top of left column, before "1 Scope". So caption first. Good. But should the caption be before h2 1—Scope? Yes.
Hmm, but actually, is the figure part of an appendix (Fig. A1)? The numbering "A1" suggests appendix. But the section numbering here is 1,2,3,4,5 (not A.1). The figure is Fig. A1. Possibly this page is from an appendix but the section numbers are 1-5? That's odd. Actually maybe the page is "222R-2" and the figure A1 belongs to an appendix but the text sections 1-5 are the main body? Hmm, but the figure is referenced in 3.2.2 ("shown in Fig. A1"). So Fig. A1 is referenced in section 3.2.2. The figure is placed at top of left column. The caption "Fig. A1—Soxhlet extraction apparatus". OK whatever, I just place caption where it visually is (top of left column = start of reading order). Good.
Now let's write the merged 2.1 paragraph carefully:
"2.1—Water-soluble chloride, when present in sufficient amounts, may initiate or accelerate the corrosion of metals such as steel embedded in or contacting a cement system such as mortar, grout, or concrete. Other test methods exist for the determination of water-soluble chloride in a cement system.* However, some aggregates contain a considerable amount of chloride that is bound in the aggregate and is not available for the corrosion reaction. The test method described in ASTM C 1218 measures a portion of the chloride contained in these aggregates. However, the amount of chloride measured is very dependent on the degree of fineness to which the aggregates are ground during sample preparation.† The problem with the ASTM C 1218 test method is therefore twofold: the test measures chlorides that are not generally available for the corrosion reaction, and the test gives widely variable results. The test method described herein should be used when chloride-bearing aggregates influence the results obtained using ASTM C 1218."
Good.
2.2: "2.2—Sulfides are known to interfere with the determination of chloride content. Blast-furnace slag aggregates and cement contain sulfide sulfur in concentrations high enough to cause significant interference and produce erroneous test results. Treatment with hydrogen peroxide, as discussed in ASTM C 114,‡ shall be used to eliminate such interference."
3.1: "3.1—Sampling equipment: The apparatus required for obtaining samples by coring or sawing is described in ASTM C 42.§ Sampling by drilling is not applicable for this test and shall not be used."
3.2: "3.2—Sample processing apparatus:"
3.2.1: "3.2.1 Samples too large to fit in the sample holder of the Soxhlet shall be reduced in size by means of a jaw crusher or by hammering."
3.2.2: "3.2.2 Extract chlorides from the sample using a Soxhlet extractor, a schematic of which is shown in Fig. A1. The Soxhlet extractor consists of a heater, a lower flask to hold water, the sample compartment, and a condenser. The extractor contains approximately 100 ml of distilled water in the lower flask. Heat is applied to this flask; vapor from the boiling water passes to the condenser; and the condensate collects in the sample compartment. The sample is contained in a porous holder and the hot condensate collects around the sample. When the condensate reaches a critical height, the liquid is siphoned back into the lower flask and the process repeats. The nonvolatile components extracted from the sample accumulate in the lower flask, while each extraction involves fresh hot distillate. The heat input shall be sufficient to give an extraction cycle about every 20 min. For convenience, suitable commercial equipment is available.**"
4.1: "4.1—The reagents required for the chloride determination are given in the test method for chloride of ASTM C 114."
5.1: "5.1—Reduce the size of a minimum 300 gm sample as specified in Section 6, and divide this sample to a minimum 30 gm representative sample for use in the chloride" (continues to next page, so it's incomplete; I'll end with "chloride" as in image. The image ends with "for use in the chloride" at bottom of right column. Yes incomplete. I'll keep as is.)
Footnotes:
small-note: "*ASTM Standard Test Method C 1218-92, Standard Test Method for Water-Soluble Chloride in Mortar and Concrete."
small-note: "†For more information see "The Determination of the Chloride Content of Concrete" by Brian B. Hope, John A. Page, and John S. Poland, Cement and Concrete Research, V. 15, No. 5, Pergamon Press, New York, Sept. 1985, pp. 863-870."
small-note: "‡ASTM Test Method C 114, "Test Methods for Chemical Analysis of Hydraulic Cement.""
small-note: "§ASTM C 42, "Test Method for Obtaining and Testing Drilled Cores and Sawn Beams of Concrete.""
small-note: "**Suitable Soxhlet extraction equipment is available from Fisher Scientific (Catalog No. 09-551A) and other manufacturers."
For the journal name "Cement and Concrete Research" it's italic in original. I'll wrap in em.
Now check forbidden things: no style, no nbsp, no markdown fences, no h1, no div, etc. Good.
Check the figure caption: should I use em for the whole caption? In original "Fig. A1—Soxhlet extraction apparatus" is italic. Yes em.
Now about the leading markers in footnotes: I'll keep them as plain text (e.g., "*ASTM..."). That's fine.
Let me double check there's no double spaces. I'll write normally.
One more: the heading "1—Scope" uses an em dash "—". Keep.
Now, should the figure caption be align-center? In the image it's left-aligned under the figure (the figure is centered in the column but the caption text starts at left margin). I'll leave as normal p.
Let me also consider: the rule "每页输出必须以
开头（页面内标题）". My output starts with the figure caption
, not h2! That violates rule 9: "每页输出必须以
开头".
Hmm. The page's first content visually is the figure (top of left column). But the rule says output must start with h2. So I need to start with an h2. The first heading on the page is "1—Scope". But the figure caption comes before it visually.
This is a conflict. The rule is highest priority: "每页输出必须以
开头". So I must start with h2. Therefore I should put the figure caption after the h2? Or maybe the figure caption should be placed where it logically belongs. Since the figure is referenced in 3.2.2, maybe I should place the figure caption near there? But visually it's at top.
Given the hard rule, I'll start with
1—Scope
and place the figure caption... where? Options:
- Place figure caption right after the h2 1—Scope? That's weird.
- Place figure caption at the end (before footnotes)?
- Place figure caption before 3.2.2 where it's referenced?
Hmm. The rule says output must start with h2. The figure caption is not a heading. So I cannot start with it. I think the cleanest: start with h2 1—Scope, and place the figure caption as a standalone p somewhere. Since the figure is at the top of the left column but the rule forces h2 first, I'll place the figure caption immediately after the h2 1—Scope heading? That doesn't match reading order either.
Alternatively, maybe I should treat the figure caption as belonging to the figure which is at top, but since I must start with h2, I'll put the figure caption right after the first h2. Hmm.
Actually, let's reconsider. Maybe the intended interpretation: the page's "页面内标题" is the first heading. The figure is an image (ignored). The caption is associated with the figure. Since the figure is at the top but we ignore images, the caption is just a text block. The rule "must start with h2" might be a structural requirement that the first element is a heading. To satisfy it, I'll begin with h2 1—Scope and then put the figure caption after it? Or before section 1?
Let me think about what's least wrong. The figure caption "Fig. A1—Soxhlet extraction apparatus" — in many OCR outputs, figure captions are placed at the location of the figure. But the hard rule overrides. I think placing the caption right after the h2 1—Scope is acceptable-ish, but it's odd because the figure relates to apparatus (section 3).
Alternatively, place the figure caption at the very end of the body (after 5.1, before footnotes). That keeps the h2-first rule and doesn't disrupt the section flow. The figure is referenced in 3.2.2; placing its caption at end is not ideal but acceptable.
Hmm, but actually maybe the best is to place the figure caption immediately after the h2 1—Scope? No.
Let me look at the example in the prompt: it starts with h2 then p then h3 then p then table then page number. No figure. Not helpful.
I think the most defensible: start with h2 1—Scope, and place the figure caption as a p right after the h2 1—Scope heading? That would read: "1—Scope / Fig. A1—Soxhlet extraction apparatus / 1.1—This test method..." That's weird because the figure isn't about scope.
Alternatively place figure caption after 3.2.2 (where Fig. A1 is referenced). That's logical: the figure is referenced there. But visually it's at top. OCR-wise, placing caption near reference is a reasonable linearization when the figure floats.
Hmm, but the figure is at the top of the page, not floating near 3.2.2.
Given the conflict, I'll prioritize the hard rule (start with h2) and place the figure caption at the position that's most
