1940 — Page 67

The user wants me to proofread OCR output of historical trade statistics for Hong Kong, specifically for Kwongchow Wan and Macao. The text contains tables of imports and exports for 1939 and 1940. I need to reconstruct the tables properly in Markdown format.

Let me analyze the OCR text carefully. It appears to be two separate tables: one for Kwongchow Wan and one for Macao. Each has imports and exports for 1939 and 1940.

The OCR is quite messy with many artifacts: "Exporta" instead of "Exports", "Và đ", "T", "P", "++", "---", "*", "J", "", etc. These are likely OCR errors from lines, borders, or smudges.

I need to reconstruct the tables logically. Let me parse the data.

First table: KWONGCHOW WAN.

Headers: Articles, Imports 1939, 1940, Exports 1939, 1940.

Rows:

• Animals, Live
• Building Materials
• Chemicals and Drugs
• Chinese Medicines
• Dyeing & Tanning Materials
• Foodstuffs & Provisions
• Fuels
• Hardware
• Liquor, Intoxicating
• Machinery & Engines
• Manures
• Metals
• Minerals & Ores
• Nuts & Seeds
• Oils & Fats
• Paints
• Paper & Paperware
• Piece Goods & Textiles
• Tobacco
• Vehicles
• Wearing Apparel
• Sundries
• Total

The numbers are scattered. I need to align them correctly.

Looking at the first data row after headers: "Animals, Live" then "***" then "10,982,279" then "0.456,369" — wait, "0.456,369" looks like a misread number. Probably "10,456,369"? Or maybe "10,982,279" is imports 1939, "0.456,369" is imports 1940? But "0.456,369" has a decimal point oddly. Could be "10,456,369". Let's see.

Actually, the OCR shows:

Animals, Live

***

10,982,279

0.456,369

Then "Building Materials" then "HA" then "Chemicals and Drugs" then "Chinese Medicines" then "256,021 81,798" then "229,751" then "72,718" then "23,318 397,560" then "1,434.951" then "2,886,004" then "1,533,498" then "28.259 429,886 1,011,093" then "Dyeing & Tanning Materials" then "Foodstuffs & Provisions" then "Fuels" then "Hardware" then "Liquor, Intoxicating" then "Machinery & Engines..." then "Manures..." then "Mietala..." then "Minerals & Dres..." then "Nuts & Seeds" then "255.774 3,947,594" then "955,215" then "569,146" then "3,573,076" then "1,738,390" then "59,145 3,940,329" then "770.014" then "639,351" then "467" then "447" then "+" then "" then "4,021" then "9,639" then "286,875" then "107,067" then "45,901" then "69,117" then "+" then "10.005" then "50" then "85,571" then "25,232" then "3,000" then "3,239" then "208.143" then "313,375" then "22,607" then "18,750" then "596.222" then "567,570" then "Và đ" then "58,830" then "624,168" then "78" then "190,503" then "117,835" then "135,027" then "95,421" then "Oils & Fate" then "Paints" then "8.710,780" then "12.409,202" then "2,742,176" then "2,868,805" then "T" then "" then "Paper & Paperware" then "Piece Goods & Textiles" then "Tcbacco" then "Vehicles" then "Weaning Apparel..." then "166,384" then "242,155" then "45,605 529,192" then "41,550" then "59,915" then "53,699" then "27,188,667" then "528,587 26,483,047" then "P" then "75,620" then "15,599" then "1,240,900" then "1,672,785" then "14" then "13,933" then "641,739" then "494.179" then "1,547" then "1,425,808" then "1,451,361" then "Sundries..." then "***" then "4,356,138" then "5,981,477" then "2,892,088" then "2,561,949" then "Total" then ":" then "28,356,970" then "37,275,842" then "$2,286,217" then "42,649,692"

This is extremely messy. The OCR has lost column alignment. I need to infer the correct numbers for each article for imports 1939, imports 1940, exports 1939, exports 1940.

Given the total row: Total imports 1939: 28,356,970; imports 1940: 37,275,842; exports 1939: $2,286,217 (maybe 2,286,217); exports 1940: 42,649,692.

But wait, the total line shows "28,356,970 37,275,842 $2,286,217 42,649,692". That's four numbers. So imports 1939, imports 1940, exports 1939, exports 1940.

Now I need to distribute the numbers across rows. This is a puzzle. Since the OCR is so garbled, I might need to make a best-effort reconstruction. However, the instruction says: "Preserve original content — do not add or remove substantive words. Do not rephrase, rewrite, or change the author's style, tone, or word choice. Only correct what is clearly wrong." But for tables, I need to reconstruct the table structure. The numbers are the content. I must not invent numbers. But the OCR has scrambled them. I need to try to align them as they appear in sequence, assuming the OCR read row by row, but the columns are interleaved.

Let me think: The OCR likely read the table in a linear fashion, maybe column by column? Or row by row but with line breaks causing misordering.

Given the complexity, perhaps the best is to present the data as a cleaned-up table with the numbers as they appear in the OCR sequence, but that would be meaningless. The user expects a proofread version of the OCR output, meaning I should correct OCR errors and format as a proper table. But I cannot guess the correct numbers if they are jumbled.

Wait, the prompt says: "You are a professional editor specializing in proofreading OCR output of historical records relating to Hong Kong." So I am to proofread the OCR text, not necessarily reconstruct the original table perfectly if the OCR is too damaged. But I should do my best to reconstruct the table based on the visible data.

Maybe the OCR text actually contains the table in a somewhat sequential order: first the header row, then each row's data in order: article name, import 1939, import 1940, export 1939, export 1940. But the OCR has inserted many artifacts and line breaks.

Let me try to parse sequentially, ignoring artifacts like "*", "HA", "T", "P", "Và đ", "+", "---", "J", "", "++", etc.

I'll go through the text token by token (numbers and article names) and try to assign columns.

First, the header: "Articles (8 17) KWONGCHOW WAN. Imports Exporta 1939. 1940. $ 1939. 1940." So columns: Articles, Imports 1939, Imports 1940, Exports 1939, Exports 1940.

Now the data rows. The OCR then lists:

"Animals, Live" -> article

"***" -> artifact

"10,982,279" -> number

"0.456,369" -> number (maybe 10,456,369? but it's separate)

"Building Materials" -> article

"HA" -> artifact

"Chemicals and Drugs" -> article

"Chinese Medicines" -> article

"256,021 81,798" -> two numbers

"229,751" -> number

"72,718" -> number

"23,318 397,560" -> two numbers

"1,434.951" -> number (note dot as thousand separator)

"2,886,004" -> number

"1,533,498" -> number

"28.259 429,886 1,011,093" -> three numbers

"Dyeing & Tanning Materials" -> article

"Foodstuffs & Provisions" -> article

"Fuels" -> article

"Hardware" -> article

"Liquor, Intoxicating" -> article

"Machinery & Engines..." -> article

"Manures..." -> article

"Mietala..." -> article (maybe "Metals"?)

"Minerals & Dres..." -> article (maybe "Minerals & Ores")

"Nuts & Seeds" -> article

"255.774 3,947,594" -> two numbers

"955,215" -> number

"569,146" -> number

"3,573,076" -> number

"1,738,390" -> number

"59,145 3,940,329" -> two numbers

"770.014" -> number

"639,351" -> number

"467" -> number

"447" -> number

"+" -> artifact

"***" -> artifact

"4,021" -> number

"9,639" -> number

"286,875" -> number

"107,067" -> number

"45,901" -> number

"69,117" -> number

"+" -> artifact

"10.005" -> number

"50" -> number

"85,571" -> number

"25,232" -> number

"3,000" -> number

"3,239" -> number

"208.143" -> number

"313,375" -> number

"22,607" -> number

"18,750" -> number

"596.222" -> number

"567,570" -> number

"Và đ" -> artifact

"58,830" -> number

"624,168" -> number

"78" -> number

"190,503" -> number

"117,835" -> number

"135,027" -> number

"95,421" -> number

"Oils & Fate" -> article (Oils & Fats)

"Paints" -> article

"8.710,780" -> number

"12.409,202" -> number

"2,742,176" -> number

"2,868,805" -> number

"T" -> artifact

"***" -> artifact

"Paper & Paperware" -> article

"Piece Goods & Textiles" -> article

"Tcbacco" -> article (Tobacco)

"Vehicles" -> article

"Weaning Apparel..." -> article (Wearing Apparel)

"166,384" -> number

"242,155" -> number

"45,605 529,192" -> two numbers

"41,550" -> number

"59,915" -> number

"53,699" -> number

"27,188,667" -> number

"528,587 26,483,047" -> two numbers

"P" -> artifact

"75,620" -> number

"15,599" -> number

"1,240,900" -> number

"1,672,785" -> number

"14" -> number

"13,933" -> number

"641,739" -> number

"494.179" -> number

"1,547" -> number

"1,425,808" -> number

"1,451,361" -> number

"Sundries..." -> article

"***" -> artifact

"4,356,138" -> number

"5,981,477" -> number

"2,892,088" -> number

"2,561,949" -> number

"Total" -> article

":" -> artifact

"28,356,970" -> number

"37,275,842" -> number

"$2,286,217" -> number

"42,649,692" -> number

Then second table: "Articles Animals, Live Building Materials --- Chemicals and Drugs Chinese Medicines Dyeing & Tanning Materials Foodstuffs & Provisions Fuels Hardware Liquor, Intoxicating Machinery & Engines Manures... Metals Minerals & Ores Nuts & Seeds +14 ... J --- Oils & Fate Painta Paper & Paperware ++ Piece Goods & Textiles Tobacco * Valncles Wearing Apparel... Sundries... Total MACAO. Importa Exports 1939. 1940. 1939. 1940. 50,960 97,310 349,154 420,483 209,451 87,959 736,934 849,786 27,080 34,562 991,284 1,481,584 3,998,078 733,006 902,218 990,984 15.440 49,764 180,750 290,478 7,923,161 6,410,823 15,025,825 17,533,440 29.890 203,435 587.870 513,421 138.547 64,893 235,154 - 258,197 611,634 896,375 611,055 698,549 --- 38,648 27,925 221.872 543,890 + --- -- 13,092 126,120 904,517 485,647 432,442 253.309 1,096,302 1,140,089 781,120 977,443 791 --- 158,334 285,617 2.885,278 0,002 1,711,133 737,787 3,622.742 6,222,622 6,589,958 5,833 12.843 141.690 137,620 * --- 705,885 387,184 1.309.602 1,741,852 7,773,482 2,317,753 4.773,259 4,988.674 506,858 279.761 2,895,955 4,250,129TM 1,870 16.209 296,667 JO 459.835 777,815 500 067 8.681,511 7.568.958 429,450 4,158.605 332,209 5,254,719 32,872,239 24.600.997 45,098,953 50,855,880 €5"

This is similarly messy.

Given the difficulty, perhaps the expected output is to present the tables as they are but cleaned of artifacts and formatted as Markdown tables with the numbers in the order they appear, assuming each row has four numbers. But the numbers don't align per row.

Alternatively, maybe the OCR output is actually two separate tables that were side by side? The first table might be for Kwongchow Wan, the second for Macao. Each has imports and exports for 1939 and 1940.

I think the best approach is to reconstruct each table by matching article names to the numbers that follow, using the total row as a check. But that's a huge data entry task and may be error-prone.

Given the instructions: "If data is clearly in table format, reconstruct the table using Markdown table syntax." So I must produce Markdown tables.

I'll attempt to parse the first table by assuming the OCR read row by row, but the line breaks cause each number to appear on a new line. However, the article names are also on separate lines. Let's try to group: each article row should have 4 numbers. Count the numbers between article names.

List article names in order as they appear:

1. Animals, Live
2. Building Materials
3. Chemicals and Drugs
4. Chinese Medicines
5. Dyeing & Tanning Materials
6. Foodstuffs & Provisions
7. Fuels
8. Hardware
9. Liquor, Intoxicating
10. Machinery & Engines...
11. Manures...
12. Mietala... (likely Metals)
13. Minerals & Dres... (likely Minerals & Ores)
14. Nuts & Seeds
15. Oils & Fate (Oils & Fats)
16. Paints
17. Paper & Paperware
18. Piece Goods & Textiles
19. Tcbacco (Tobacco)
20. Vehicles
21. Weaning Apparel... (Wearing Apparel)
22. Sundries...
23. Total

That's 23 rows including total.

Now count numbers in the sequence between start and total. I'll list all numbers in order as they appear in the OCR text (ignoring artifacts). I'll copy the number tokens:

1. 10,982,279
2. 0.456,369
3. 256,021
4. 81,798
5. 229,751
6. 72,718
7. 23,318
8. 397,560
9. 1,434.951
10. 2,886,004
11. 1,533,498
12. 28.259
13. 429,886
14. 1,011,093
15. 255.774
16. 3,947,594
17. 955,215
18. 569,146
19. 3,573,076
20. 1,738,390
21. 59,145
22. 3,940,329
23. 770.014
24. 639,351
25. 467
26. 447
27. 4,021
28. 9,639
29. 286,875
30. 107,067
31. 45,901
32. 69,117
33. 10.005
34. 50
35. 85,571
36. 25,232
37. 3,000
38. 3,239
39. 208.143
40. 313,375
41. 22,607
42. 18,750
43. 596.222
44. 567,570
45. 58,830
46. 624,168
47. 78
48. 190,503
49. 117,835
50. 135,027
51. 95,421
52. 8.710,780
53. 12.409,202
54. 2,742,176
55. 2,868,805
56. 166,384
57. 242,155
58. 45,605
59. 529,192
60. 41,550
61. 59,915
62. 53,699
63. 27,188,667
64. 528,587
65. 26,483,047
66. 75,620
67. 15,599
68. 1,240,900
69. 1,672,785
70. 14
71. 13,933
72. 641,739
73. 494.179
74. 1,547
75. 1,425,808
76. 1,451,361
77. 4,356,138
78. 5,981,477
79. 2,892,088
80. 2,561,949
81. 28,356,970
82. 37,275,842
83. 2,286,217 (from $2,286,217)
84. 42,649,692

That's 84 numbers. For 23 rows, each with 4 numbers = 92 numbers. We have 84, missing 8 numbers. Some rows might have missing numbers (like "---" indicating zero or not applicable). The artifacts like "---" might represent missing data.

Also, some numbers might be combined incorrectly: e.g., "256,021 81,798" are two numbers. "23,318 397,560" two numbers. "28.259 429,886 1,011,093" three numbers. "255.774 3,947,594" two. "59,145 3,940,329" two. "45,605 529,192" two. "528,587 26,483,047" two. So the count is correct.

But we have 84 numbers for 22 data rows (excluding total) = 88 expected. Total row has 4 numbers. So 22*4=88, plus 4 =92. We have 84. So 8 numbers missing. Possibly some rows have only 3 numbers? Or some numbers are not captured.

Given the complexity, maybe the OCR output is not meant to be perfectly reconstructed; the proofreader should clean up the text and present the tables as best as possible, perhaps with placeholders for missing data.

But the instruction: "Missing text — if a word or phrase is clearly missing due to OCR damage, insert ... in its place." So for missing numbers, I could insert ....

However, the numbers are not words; they are data. But the same principle applies.

Given the time, I think the best is to produce two Markdown tables with the article names and the numbers as they appear in sequence, but aligned per row by assuming each article gets the next four numbers. But the sequence of numbers is not grouped per article; the OCR seems to have read column by column? Let's check: The first few numbers: 10,982,279 and 0.456,369 could be imports 1939 and 1940 for Animals, Live. Then Building Materials has no numbers? Then Chemicals and Drugs and Chinese Medicines have numbers: 256,021, 81,798, 229,751, 72,718? That's four numbers for two articles? Actually, Chemicals and Drugs and Chinese Medicines are two separate articles. They might each have four numbers. But we see four numbers before the next article (Dyeing & Tanning Materials). So maybe Chemicals and Drugs gets 256,021 and 81,798? And Chinese Medicines gets 229,751 and 72,718? But each article should have four numbers (imports 1939, 1940, exports 1939, 1940). So two numbers per article is not enough.

Maybe the table has only two columns for imports and two for exports? Actually, the header shows Imports 1939, 1940, Exports 1939, 1940. So four columns.

Perhaps the OCR read the table in a different order: first all imports 1939, then all imports 1940, then all exports 1939, then all exports 1940. That would explain the sequence: first column (imports 1939) for all articles, then second column, etc. But the article names are interspersed.

Look at the OCR: after "Animals, Live" we have two numbers. Then "Building Materials" with no numbers. Then "Chemicals and Drugs" and "Chinese Medicines" then four numbers. Then "Dyeing & Tanning Materials", "Foodstuffs & Provisions", "Fuels", "Hardware", "Liquor, Intoxicating", "Machinery & Engines...", "Manures...", "Mietala...", "Minerals & Dres...", "Nuts & Seeds" then many numbers. This suggests the article names are listed first, then the data columns follow. But the OCR has mixed them.

Actually, the original scan might have had the article names in a column on the left, and then four columns of numbers. The OCR might have read the article names column first, then the number columns. But the OCR output shows article names interspersed with numbers.

Given the difficulty, I'll assume the OCR text is a linear representation of the table rows, but with line breaks causing each cell to be on a new line. So I'll parse it as a sequence of cells: article, import1939, import1940, export1939, export1940, article, import1939, ... etc. But the artifacts break this.

Let's try to parse linearly, skipping artifacts, and see if we can group every 5 tokens (article + 4 numbers) but the article names sometimes have multiple words.

Better: I'll write a script mentally? Not possible. I'll do manual grouping for the first table.

I'll go through the OCR text line by line as provided in the user message. The user message shows the OCR text with line breaks. I'll copy it and analyze.

The user message:

Articles 


(8 17 ) 

KWONGCHOW WAN. 

Imports 

Exporta 

1939. 

1940. $ 

1939. 

1940. 

Animals, Live 

*** 

10,982,279 

0.456,369 

Building Materials 

HA 

Chemicals and Drugs 

Chinese Medicines 

256,021 81,798 

229,751 

72,718 

23,318 397,560 

1,434.951 

2,886,004 

1,533,498 

28.259 429,886 1,011,093 

Dyeing & Tanning Materials 

Foodstuffs & Provisions 

Fuels 

Hardware 

Liquor, Intoxicating 

Machinery & Engines... 

Manures... 

Mietala... 

Minerals & Dres... 

Nuts & Seeds 

255.774 3,947,594 

955,215 

569,146 

3,573,076 

1,738,390 

59,145 3,940,329 

770.014 

639,351 

467 

447 

+ 

*** 

4,021 

9,639 

286,875 

107,067 

45,901 

69,117 

+ 

10.005 

50 

85,571 

25,232 

3,000 

3,239 

208.143 

313,375 

22,607 

18,750 

596.222 

567,570 

Và đ 

58,830 

624,168 

78 

190,503 

117,835 

135,027 

95,421 

Oils & Fate 

Paints 

8.710,780 

12.409,202 

2,742,176 

2,868,805 

T 

*** 

Paper & Paperware 

Piece Goods & Textiles 

Tcbacco 

Vehicles 

Weaning Apparel... 

166,384 

242,155 

45,605 529,192 

41,550 

59,915 

53,699 

27,188,667 

528,587 26,483,047 

P 

75,620 

15,599 

1,240,900 

1,672,785 

14 

13,933 

641,739 

494.179 

1,547 

1,425,808 

1,451,361 

Sundries... 

*** 

4,356,138 

5,981,477 

2,892,088 

2,561,949 

Total 

: 

28,356,970 

37,275,842 

$2,286,217 

42,649,692 

Then second table.

I'll treat this as a text stream. I'll write a parser in my mind: iterate lines, skip empty lines, skip lines that are artifacts (like "*", "HA", "T", "P", "Và đ", "+", "---", "J", "", "++", ":"), and collect article names and numbers.

But article names can be multi-word. The lines that are not numbers and not artifacts are article names. Numbers lines contain numbers (with commas, dots). Some lines have two numbers separated by space.

Let's list lines with content:

1. Articles
2. (8 17 )
3. KWONGCHOW WAN.
4. Imports
5. Exporta
6. 1939.
7. 1940. $
8. 1939.
9. 1940.
10. Animals, Live
11. ***
12. 10,982,279
13. 0.456,369
14. Building Materials
15. HA
16. Chemicals and Drugs
17. Chinese Medicines
18. 256,021 81,798
19. 229,751
20. 72,718
21. 23,318 397,560
22. 1,434.951
23. 2,886,004
24. 1,533,498
25. 28.259 429,886 1,011,093
26. Dyeing & Tanning Materials
27. Foodstuffs & Provisions
28. Fuels
29. Hardware
30. Liquor, Intoxicating
31. Machinery & Engines...
32. Manures...
33. Mietala...
34. Minerals & Dres...
35. Nuts & Seeds
36. 255.774 3,947,594
37. 955,215
38. 569,146
39. 3,573,076
40. 1,738,390
41. 59,145 3,940,329
42. 770.014
43. 639,351
44. 467
45. 447
46. +
47. ***
48. 4,021
49. 9,639
50. 286,875
51. 107,067
52. 45,901
53. 69,117
54. +
55. 10.005
56. 50
57. 85,571
58. 25,232
59. 3,000
60. 3,239
61. 208.143
62. 313,375
63. 22,607
64. 18,750
65. 596.222
66. 567,570
67. Và đ
68. 58,830
69. 624,168
70. 78
71. 190,503
72. 117,835
73. 135,027
74. 95,421
75. Oils & Fate
76. Paints
77. 8.710,780
78. 12.409,202
79. 2,742,176
80. 2,868,805
81. T
82. ***
83. Paper & Paperware
84. Piece Goods & Textiles
85. Tcbacco
86. Vehicles
87. Weaning Apparel...
88. 166,384
89. 242,155
90. 45,605 529,192
91. 41,550
92. 59,915
93. 53,699
94. 27,188,667
95. 528,587 26,483,047
96. P
97. 75,620
98. 15,599
99. 1,240,900
100. 1,672,785
101. 14
102. 13,933
103. 641,739
104. 494.179
105. 1,547
106. 1,425,808
107. 1,451,361
108. Sundries...
109. ***
110. 4,356,138
111. 5,981,477
112. 2,892,088
113. 2,561,949
114. Total
115. :
116. 28,356,970
117. 37,275,842
118. $2,286,217
119. 42,649,692

Now, I need to assign each article its four numbers. The articles appear at lines: 10, 14, 16, 17, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 75, 76, 83, 84, 85, 86, 87, 108, 114. That's 23 articles (including Total). Good.

Now, the numbers appear in lines 12-13, 18-25, 36-45, 48-66, 68-74, 77-80, 88-95, 97-107, 110-113, 116-119.

But the numbers are not grouped per article. It seems the OCR read the table column by column? Let's see: After "Animals, Live" we have two numbers (lines 12,13). Then "Building Materials" with no numbers. Then "Chemicals and Drugs" and "Chinese Medicines" then a bunch of numbers (lines 18-25). Then a list of articles (lines 26-35) with no numbers. Then numbers (lines 36-45). Then artifacts, then numbers (48-66). Then "Và đ" artifact, then numbers (68-74). Then "Oils & Fate", "Paints", then numbers (77-80). Then artifacts, then articles (83-87), then numbers (88-95). Then artifact, numbers (97-107). Then "Sundries...", artifact, numbers (110-113). Then Total, numbers (116-119).

This suggests the table might have been split into multiple sections? Or the OCR read the left column (articles) first, then the data columns. But the articles are not all together; they appear in two blocks: first block lines 10-35 (14 articles), second block lines 75-87 (7 articles), then Sundries, Total. That's 22 articles plus total.

First block articles: Animals, Live; Building Materials; Chemicals and Drugs; Chinese Medicines; Dyeing & Tanning Materials; Foodstuffs & Provisions; Fuels; Hardware; Liquor, Intoxicating; Machinery & Engines...; Manures...; Mietala...; Minerals & Dres...; Nuts & Seeds. That's 14 articles.

Second block: Oils & Fate; Paints; Paper & Paperware; Piece Goods & Textiles; Tcbacco; Vehicles; Weaning Apparel... That's 7 articles.

Then Sundries... and Total.

Total 23 rows.

Now, the numbers: There are many numbers. Perhaps each block of numbers corresponds to a column for all articles. For example, lines 12-13: two numbers for first article only? But then lines 18-25: 8 numbers? Let's count numbers in each block.

Block 1 (lines 12-13): 2 numbers.

Block 2 (lines 18-25): line18 has 2, line19 1, line20 1, line21 2, line22 1, line23 1, line24 1, line25 3 => total 12 numbers.

Block 3 (lines 36-45): line36 2, line37 1, line38 1, line39 1, line40 1, line41 2, line42 1, line43 1, line44 1, line45 1 => total 12 numbers.

Block 4 (lines 48-66): many lines, each with 1 number except line50? Let's count: lines 48,49,50,51,52,53,55,56,57,58,59,60,61,62,63,64,65,66 => 18 numbers? Actually line54 is "+", line55 is "10.005", etc. So 18 numbers.

Block 5 (lines 68-74): 7 numbers.

Block 6 (lines 77-80): 4 numbers.

Block 7 (lines 88-95): line88 1, line89 1, line90 2, line91 1, line92 1, line93 1, line94 1, line95 2 => 10 numbers.

Block 8 (lines 97-107): line97 1, line98 1, line99 1, line100 1, line101 1, line102 1, line103 1, line104 1, line105 1, line106 1, line107 1 => 11 numbers.

Block 9 (lines 110-113): 4 numbers.

Block 10 (lines 116-119): 4 numbers.

Total numbers: 2+12+12+18+7+4+10+11+4+4 = 84. As before.

If each article has 4 numbers, 23 articles = 92 numbers. We have 84. So 8 missing. Could be that some articles have no data (like Building Materials) and are represented by "---" or something. In the OCR, we see "---" in the second table but not in first. In first table, "Building Materials" has "HA" after it, maybe "HA" means "N/A"? And "Mietala..." might be "Metals" with no data? But there are numbers later.

Maybe the table has only 3 columns? But header shows 4.

Given the time constraints, I think the best is to output the tables as they appear in the OCR but cleaned up: remove artifacts, fix obvious typos (e.g., "Exporta" -> "Exports", "Oils & Fate" -> "Oils & Fats", "Tcbacco" -> "Tobacco", "Weaning Apparel" -> "Wearing Apparel", "Valncles" -> "Vehicles", "Painta" -> "Paints", "Importa" -> "Imports", "Mietala" -> "Metals", "Minerals & Dres" -> "Minerals & Ores", "Và đ" remove, etc.). And format as Markdown tables with the numbers placed in columns as they appear sequentially per row, but since we can't align perfectly, we might need to present the data in a simplified way.

However, the instruction says: "Reconstruct the table using Markdown table syntax." So I must produce a table. I'll attempt to reconstruct the first table by assuming the numbers are in row-major order as they appear after the article names in the text, but the article names are not in the same order as numbers.

Wait, maybe the OCR text is actually the table in a tab-separated format but with line breaks. Let's look at the raw text: "Animals, Live 10,982,279 0.456,369 Building Materials HA Chemicals and Drugs Chinese Medicines 256,021 81,798 229,751 72,718 23,318 397,560 1,434.951 2,886,004 1,533,498 28.259 429,886 1,011,093 Dyeing & Tanning Materials Foodstuffs & Provisions Fuels Hardware Liquor, Intoxicating Machinery & Engines... Manures... Mietala... Minerals & Dres... Nuts & Seeds 255.774 3,947,594 955,215 569,146 3,573,076 1,738,390 59,145 3,940,329 770.014 639,351 467 447 + 4,021 9,639 286,875 107,067 45,901 69,117 + 10.005 50 85,571 25,232 3,000 3,239 208.143 313,375 22,607 18,750 596.222 567,570 Và đ 58,830 624,168 78 190,503 117,835 135,027 95,421 Oils & Fate Paints 8.710,780 12.409,202 2,742,176 2,868,805 T Paper & Paperware Piece Goods & Textiles Tcbacco Vehicles Weaning Apparel... 166,384 242,155 45,605 529,192 41,550 59,915 53,699 27,188,667 528,587 26,483,047 P 75,620 15,599 1,240,900 1,672,785 14 13,933 641,739 494.179 1,547 1,425,808 1,451,361 Sundries... 4,356,138 5,981,477 2,892,088 2,561,949 Total : 28,356,970 37,275,842 $2,286,217 42,649,692"

If I split by spaces, but numbers have commas. This is too messy.

Given the difficulty, perhaps the user expects me to output the text as cleaned-up paragraphs and tables, but for the tables, I can present them as they are with the numbers in the same order but in a table with columns: Article, Imports 1939, Imports 1940, Exports 1939, Exports 1940, and fill in the numbers from the total row and maybe leave others as ...? But that would be incomplete.

Alternatively, I could output the OCR text as a cleaned-up version with the tables represented in Markdown but with the data rows as they appear in the OCR, i.e., each line of the OCR becomes a row? No.

I think I need to make a judgment call. Since this is a proofreading task, the primary goal is to fix the text flow and correct OCR errors. For the tables, I should reconstruct them as best as possible. I'll attempt to align the first table by using the article list in order and the numbers in the order they appear after the article list, but there are two article lists.

Maybe the table is actually two separate tables? The header "KWONGCHOW WAN." then "Imports Exporta 1939. 1940. $ 1939. 1940." Then the data. The article names might be repeated for imports and exports? No.

Let's search for similar historical data. Kwongchow Wan (Guangzhouwan) was a French leased territory. Trade statistics might be available. But I cannot rely on external knowledge.

Given the instructions, I will produce a Markdown document with two tables. For each table, I will list the articles as they appear in the OCR (in the order they appear) and then for each article, I will take the next four numbers from the number sequence, but the numbers are not grouped per article. However, I can assume that the OCR read the table row by row, but the line breaks cause each cell to be on a new line. So if I ignore the line breaks and treat the text as a sequence of tokens (article names and numbers), I can parse it as: article, num1, num2, num3, num4, article, num1, num2, num3, num4, ... But the artifacts break this.

Let's try to tokenize the entire first table text (from "Animals, Live" to "42,649,692") by splitting on whitespace and newlines, but keeping numbers with commas together. Then filter out artifacts. Then see if the pattern emerges.

I'll do a manual tokenization of the first table text as a single string (excluding the header lines). The text from "Animals, Live" to "42,649,692" as provided in the user message (with line breaks). I'll copy that segment and mentally split.

But perhaps the user expects me to output the proofread text with the tables formatted nicely, even if the numbers are not perfectly aligned, but with the correct article names and the numbers as they appear in the OCR, but in a table with columns. Since the OCR is so damaged, I might indicate missing data with "...".

Given the time, I'll produce a cleaned-up version of the text, with the tables represented as Markdown tables with the article names and the numbers placed in the columns based on the total row and the visible numbers for each article where possible. For the first table, I'll try to match each article to four numbers by using the sequence of numbers after the article name until the next article name. But the article names are not always followed by numbers; sometimes numbers appear before the article name.

Look at the sequence: After "Animals, Live" we have two numbers. Then "Building Materials" with no numbers. Then "Chemicals and Drugs" and "Chinese Medicines" then many numbers. Then a list of articles with no numbers. Then numbers. This suggests the numbers are grouped by column, not by row.

Thus, the table might have been read column-wise: first column (Imports 1939) for all articles, then second column (Imports 1940), etc. But the article names are listed only once at the top? But they appear interspersed.

Maybe the OCR output is from a PDF with two pages? The first part might be the imports table, the second part exports? But the header says both imports and exports.

I think I need to output something. I'll do the following: For each table, I'll create a Markdown table with the columns: Article, Imports 1939, Imports 1940, Exports 1939, Exports 1940. I'll fill in the article names in the order they appear in the OCR (first block then second block). For the numbers, I'll use the total row as the last row. For the other rows, I'll put "..." for each cell, because the data is too scrambled to reliably assign. But that seems like giving up.

Alternatively, I could output the raw numbers in a single column? No.

Given the instruction "Preserve original content — do not add or remove substantive words. Do not rephrase, rewrite, or change the author's style, tone, or word choice. Only correct what is clearly wrong." The numbers are substantive content. I must not remove them. So I must include all numbers in the output. But how to present them in a table? Maybe the table is not meant to be perfectly reconstructed; the proofread text could present the data as a list of articles with their numbers in the order they appear, but that's not a table.

The instruction says: "If data is clearly in table format, reconstruct the table using Markdown table syntax." So I must attempt a table.

I'll try to reconstruct the first table by assuming the numbers are in row-major order for the 22 articles (excluding total) but the article names are listed in two groups. Perhaps the first group of articles (14) correspond to the first 14 rows, and the second group (7) correspond to rows 15-21, then Sundries row 22, Total row 23. The numbers might be in column-major order: first all imports 1939 for 22 articles, then all imports 1940, then all exports 1939, then all exports 1940. But we have 84 numbers, which is 214 = 84. That matches! 21 articles (excluding total) 4 = 84. And total row has 4 numbers separately. So the 84 numbers correspond to the 21 data rows (articles) in column-major order: first column (imports 1939) for 21 articles, then second column (imports 1940) for 21 articles, then third column (exports 1939) for 21 articles, then fourth column (exports 1940) for 21 articles. That would be 21*4=84 numbers. And the total row is separate.

Let's test this hypothesis. If the numbers are in column-major order, then the first 21 numbers are imports 1939 for each article in order. The next 21 are imports 1940, next 21 exports 1939, last 21 exports 1940.

We have 84 numbers. Let's list them in order as they appear in the OCR (the 84 numbers I listed earlier). Then we can assign them to articles in the order the articles appear in the OCR (the 21 articles before total). The articles in order (excluding total) are:

1. Animals, Live
2. Building Materials
3. Chemicals and Drugs
4. Chinese Medicines
5. Dyeing & Tanning Materials
6. Foodstuffs & Provisions
7. Fuels
8. Hardware
9. Liquor, Intoxicating
10. Machinery & Engines...
11. Manures...
12. Mietala... (Metals)
13. Minerals & Dres... (Minerals & Ores)
14. Nuts & Seeds
15. Oils & Fate (Oils & Fats)
16. Paints
17. Paper & Paperware
18. Piece Goods & Textiles
19. Tcbacco (Tobacco)
20. Vehicles
21. Weaning Apparel... (Wearing Apparel)
22. Sundries... (but wait, that's 22? Let's count: I have 22 articles before total? Actually, I listed 23 including total. Excluding total, 22 articles. But 22*4=88, not 84. So maybe Sundries is not included in the 21? Or Building Materials has no data? Let's recount articles from the OCR lines:

Article lines (non-number, non-artifact, non-header):

1. Animals, Live
2. Building Materials
3. Chemicals and Drugs
4. Chinese Medicines
5. Dyeing & Tanning Materials
6. Foodstuffs & Provisions
7. Fuels
8. Hardware
9. Liquor, Intoxicating
10. Machinery & Engines...
11. Manures...
12. Mietala...
13. Minerals & Dres...
14. Nuts & Seeds
15. Oils & Fate
16. Paints
17. Paper & Paperware
18. Piece Goods & Textiles
19. Tcbacco
20. Vehicles
21. Weaning Apparel...
22. Sundries...

That's 22 articles. Plus Total = 23 rows. But we have 84 numbers for data rows. 84/4 = 21. So one article might not have data (maybe Building Materials has no data, or Sundries is part of the total? But Sundries has numbers after it: lines 110-113 are four numbers. Those could be the four numbers for Sundries. But in the column-major scheme, Sundries would be one of the 21. Let's see: If we have 21 articles with data, which one is missing? Building Materials has no numbers near it. Also, "Mietala..." and "Minerals & Dres..." might be sub-items? But they are listed as separate articles.

Maybe the article "Building Materials" is not a data row but a header? But it's listed as an article.

Let's check the second table: it has "Building Materials" again. Materials" as an article with numbers.

In the first table, after "Building Materials" there is "HA" which might be "N/A". So perhaps Building Materials has no data, so it's not included in the 21 data rows. That would give 21 data rows. Let's assume the 21 data rows are the 22 articles minus Building Materials. Then the order of data rows would be the other 21 articles in the same order.

But then the numbers in column-major order: first 21 numbers = imports 1939 for each of the 21 articles. Let's list the first 21 numbers from my numbered list:

1. 10,982,279
2. 0.456,369
3. 256,021
4. 81,798
5. 229,751
6. 72,718
7. 23,318
8. 397,560
9. 1,434.951
10. 2,886,004
11. 1,533,498
12. 28.259
13. 429,886
14. 1,011,093
15. 255.774
16. 3,947,594
17. 955,215
18. 569,146
19. 3,573,076
20. 1,738,390
21. 59,145

That's 21 numbers. Then next 21 numbers (imports 1940) would be numbers 22-42:

1. 3,940,329
2. 770.014
3. 639,351
4. 467
5. 447
6. 4,021
7. 9,639
8. 286,875
9. 107,067
10. 45,901
11. 69,117
12. 10.005
13. 50
14. 85,571
15. 25,232
16. 3,000
17. 3,239
18. 208.143
19. 313,375
20. 22,607
21. 18,750

Next 21 (exports 1939) numbers 43-63:

1. 596.222
2. 567,570
3. 58,830
4. 624,168
5. 78
6. 190,503
7. 117,835
8. 135,027
9. 95,421
10. 8.710,780
11. 12.409,202
12. 2,742,176
13. 2,868,805
14. 166,384
15. 242,155
16. 45,605
17. 529,192
18. 41,550
19. 59,915
20. 53,699
21. 27,188,667

Next 21 (exports 1940) numbers 64-84:

1. 528,587
2. 26,483,047
3. 75,620
4. 15,599
5. 1,240,900
6. 1,672,785
7. 14
8. 13,933
9. 641,739
10. 494.179
11. 1,547
12. 1,425,808
13. 1,451,361
14. 4,356,138
15. 5,981,477
16. 2,892,088
17. 2,561,949
18. 28,356,970
19. 37,275,842
20. 2,286,217
21. 42,649,692

But wait, numbers 81-84 are the total row numbers. In my list, the total row numbers are the last four. But in the column-major scheme, the last 21 numbers should be exports 1940 for the 21 articles. But we have only 21 numbers for exports 1940? Actually, numbers 64-84 are 21 numbers? 64 to 84 inclusive is 21 numbers. But 81-84 are the total row. So the total row numbers are mixed in? That suggests the total row is not separate; it's part of the column-major data? But the total row is a separate row. In column-major, the total would be an extra row, making 22 rows. But we have 84 numbers = 21*4. So the total row is not included in the 84. But in my list, the last four numbers (81-84) are the total row. But they appear as numbers 81-84 in the sequence. That would mean the exports 1940 column has only 17 numbers? Let's recount: I have 84 numbers total. If the first 21 are imports 1939, next 21 imports 1940, next 21 exports 1939, that's 63. Then exports 1940 should be 21 numbers, making 84. But my numbers 64-84 are 21 numbers. However, numbers 81-84 are the total row. So the total row numbers are included in the exports 1940 column? That can't be.

Let's check the OCR: after "Sundries... ** 4,356,138 5,981,477 2,892,088 2,561,949 Total : 28,356,970 37,275,842 $2,286,217 42,649,692". So the four numbers after Sundries are 4,356,138; 5,981,477; 2,892,088; 2,561,949. Then Total row has four numbers. So there are 4 numbers for Sundries, and 4 for Total. That's 8 numbers at the end. In my numbered list, I have 4 numbers for Sundries (110-113) and 4 for Total (116-119). That's 8 numbers. But in my sequential list of 84 numbers, I included both. So the 84 numbers include Sundries and Total. But Sundries is an article, Total is not. So there are 22 articles (including Sundries) plus Total. 22 articles 4 = 88 numbers. We have 84. So 4 numbers missing. Perhaps Building Materials has no numbers (0), and they are omitted. And maybe one other article has missing numbers.

Given the confusion, I think the column-major hypothesis is plausible but the total row is separate. The 84 numbers might be for 21 articles (excluding Building Materials and maybe one other). But Sundries is included. Let's assume the 21 articles are all except Building Materials. Then the order of articles for data rows is the 21 articles in the order they appear, skipping Building Materials. That order:

1. Animals, Live
2. Chemicals and Drugs
3. Chinese Medicines
4. Dyeing & Tanning Materials
5. Foodstuffs & Provisions
6. Fuels
7. Hardware
8. Liquor, Intoxicating
9. Machinery & Engines...
10. Manures...
11. Mietala... (Metals)
12. Minerals & Dres... (Minerals & Ores)
13. Nuts & Seeds
14. Oils & Fate
15. Paints
16. Paper & Paperware
17. Piece Goods & Textiles
18. Tcbacco (Tobacco)
19. Vehicles
20. Weaning Apparel... (Wearing Apparel)
21. Sundries...

That's 21 articles. Good. Then the numbers in column-major order: first 21 = imports 1939, next 21 = imports 1940, next 21 = exports 1939, next 21 = exports 1940. And the total row is separate.

Now, does the sequence of 84 numbers match this? We have 84 numbers. Let's split them into 4 groups of 21.

Group1 (imports 1939): numbers 1-21 from my list.

Group2 (imports 1940): numbers 22-42.

Group3 (exports 1939): numbers 43-63.

Group4 (exports 1940): numbers 64-84.

But my list has 84 numbers, with the last four being the total row. However, in the OCR, the total row appears after Sundries. In the number sequence, the Sundries numbers appear before the total row. In my list, numbers 77-80 are Sundries? Let's see: numbers 77-80 are 4,356,138; 5,981,477; 2,892,088; 2,561,949. Those are the four numbers for Sundries. Then numbers 81-84 are total. So in the 84-number sequence, the last 8 numbers are Sundries (4) and Total (4). But in the column-major scheme, the last group (exports 1940) should have 21 numbers, but we have only 17 numbers before Sundries? Let's calculate: Group4 should be numbers 64-84 (21 numbers). But numbers 64-84 include Sundries and Total. That would mean the exports 1940 for the last few articles are missing, and instead we have Sundries and Total. This is messy.

Maybe the table is not column-major. Perhaps the OCR read the table row by row, but the rows are split across lines. The fact that the article names appear in two blocks suggests the OCR read the first column (articles) for the first half of the table, then the data columns for the first half, then the articles for the second half, then data for the second half. This is common in OCR of multi-page tables or tables with repeated headers.

Given the time, I will produce a proofread version that presents the data as two tables with the article names and the numbers as they appear in the OCR, but in a row-wise format using the sequence of numbers as they appear after each article name in the text, but since it's not consistent, I'll do a simplified version: I'll create a table for Kwongchow Wan with the articles listed in the order they appear in the OCR (first block then second block) and for each article, I'll put the numbers that appear closest to it in the text. But that's subjective.

Given the instructions, maybe the user expects me to clean up the OCR text and output it as Markdown with the tables represented as they are, but with corrected spelling and formatting. For the tables, I can output them as Markdown tables with the header and then each row as a series of cells, but if the data is garbled, I can put the raw numbers in a single cell? No.

I think I need to make a decision. I'll output the text with the tables reconstructed as best as I can, using the column-major hypothesis for the first table, because it yields a complete table with 21 rows and 4 columns, and the total row separate. I'll use the article list (excluding Building Materials) in the order they appear. I'll take the 84 numbers and split into 4 columns of 21. Then I'll create the table. For the total row, I'll use the four total numbers.

But I must ensure the numbers are correctly assigned. Let's do that systematically.

First, list the 21 articles in order (as they appear in the OCR, skipping Building Materials):

1. Animals, Live
2. Chemicals and Drugs
3. Chinese Medicines
4. Dyeing & Tanning Materials
5. Foodstuffs & Provisions
6. Fuels
7. Hardware
8. Liquor, Intoxicating
9. Machinery & Engines
10. Manures
11. Metals (Mietala)
12. Minerals & Ores (Minerals & Dres)
13. Nuts & Seeds
14. Oils & Fats (Oils & Fate)
15. Paints
16. Paper & Paperware
17. Piece Goods & Textiles
18. Tobacco (Tcbacco)
19. Vehicles
20. Wearing Apparel (Weaning Apparel)
21. Sundries

Now, the 84 numbers in order as they appear in the OCR (from my numbered list 1-84). But note: my numbered list includes numbers from the entire OCR, including those that appear before the second block of articles. The order of numbers in the OCR is the order they appear in the text. That order is likely the column-major order: first all imports 1939 for all 21 articles, then all imports 1940, then all exports 1939, then all exports 1940. But the article names are not in that order in the text; the text has article names interspersed. However, the numbers appear in blocks: first block (lines 12-25) has 14 numbers? Actually, we need to extract the numbers in the exact sequence they appear in the OCR text, ignoring artifacts and article names. That sequence is what I listed as 84 numbers. That sequence should be the column-major data if the OCR read the table column by column. But the article names are also in the text, but they appear in row order. The OCR output mixes them. But the numbers themselves are in a sequence. If the OCR processed the image left-to-right, top-to-bottom, it would read the first column (articles) then the second column (imports 1939) then third column (imports 1940) etc. But the article column would be read first, giving all article names. Then the numbers columns. However, the OCR output shows article names, then some numbers, then more article names, then more numbers. This suggests the table might be split across two pages or sections.

Given the difficulty, I'll assume the 84 numbers are in row-major order for the 21 articles? 21*4=84. So if I take the 84 numbers in order and group them into 21 groups of 4, each group corresponds to an article in the order the articles appear in the OCR (first block then second block). But the articles appear in two blocks: first block 14 articles (including Building Materials), second block 7 articles, then Sundries. That's 22 articles. If I skip Building Materials, I have 21 articles. The order of articles in the OCR (excluding Building Materials) is: Animals, Live; Chemicals and Drugs; Chinese Medicines; Dyeing & Tanning Materials; Foodstuffs & Provisions; Fuels; Hardware; Liquor, Intoxicating; Machinery & Engines...; Manures...; Mietala...; Minerals & Dres...; Nuts & Seeds; Oils & Fate; Paints; Paper & Paperware; Piece Goods & Textiles; Tcbacco; Vehicles; Weaning Apparel...; Sundries... That's 21. Good.

Now, the numbers in the OCR appear in a sequence. If the OCR read the table row by row, the numbers for each article would appear together. But they don't; they are scattered. However, if we take the entire sequence of numbers (84) and divide into 21 groups of 4 sequentially, we can assign each group to an article in that order. This assumes the numbers are in row-major order in the OCR stream. But are they? The OCR stream has article names interspersed. The numbers appear in blocks that don't align with article names. But if we ignore the article names and just take the numbers in the order they appear, we get a sequence. That sequence might be the row-major order of the table if the OCR read the table row by row but the article names were recognized separately? Unlikely.

Let's test: Take the first 4 numbers: 10,982,279; 0.456,369; 256,021; 81,798. Assign to Animals, Live. Next 4: 229,751; 72,718; 23,318; 397,560 -> Chemicals and Drugs. Next 4: 1,434.951; 2,886,004; 1,533,498; 28.259 -> Chinese Medicines. Next 4: 429,886; 1,011,093; 255.774; 3,947,594 -> Dyeing & Tanning Materials. Next 4: 955,215; 569,146; 3,573,076; 1,738,390 -> Foodstuffs & Provisions. Next 4: 59,145; 3,940,329; 770.014; 639,351 -> Fuels. Next 4: 467; 447; 4,021; 9,639 -> Hardware. Next 4: 286,875; 107,067; 45,901; 69,117 -> Liquor, Intoxicating. Next 4: 10.005; 50; 85,571; 25,232 -> Machinery & Engines. Next 4: 3,000; 3,239; 208.143; 313,375 -> Manures. Next 4: 22,607; 18,750; 596.222; 567,570 -> Metals. Next 4: 58,830; 624,168; 78; 190,503 -> Minerals & Ores. Next 4: 117,835; 135,027; 95,421; 8.710,780 -> Nuts & Seeds. Next 4: 12.409,202; 2,742,176; 2,868,805; 166,384 -> Oils & Fats. Next 4: 242,155; 45,605; 529,192; 41,550 -> Paints. Next 4: 59,915; 53,699; 27,188,667; 528,587 -> Paper & Paperware. Next 4: 26,483,047; 75,620; 15,599; 1,240,900 -> Piece Goods & Textiles. Next 4: 1,672,785; 14; 13,933; 641,739 -> Tobacco. Next 4: 494.179; 1,547; 1,425,808; 1,451,361 -> Vehicles. Next 4: 4,356,138; 5,981,477; 2,892,088; 2,561,949 -> Wearing Apparel. But we have 21 articles, so the last article Sundries would get the next 4 numbers, but we have only the total row numbers left: 28,356,970; 37,275,842; 2,286,217; 42,649,692. That would assign total numbers to Sundries, which is wrong. And Sundries actually has its own numbers (4,356,138 etc.) which we assigned to Wearing Apparel. So this grouping is off by one.

If we instead start grouping after the first two numbers? Because the first article "Animals, Live" has only two numbers in the OCR before the next article? But in the row-major hypothesis, each article should have 4 numbers. The OCR shows two numbers after Animals, Live, then Building Materials, then Chemicals and Drugs, etc. So the numbers are not grouped per article.

Given the total row numbers are at the end, and Sundries has four numbers before Total, it's likely that the last 8 numbers are Sundries (4) and Total (4). In my 84-number list, the last 8 are: 4,356,138; 5,981,477; 2,892,088; 2,561,949; 28,356,970; 37,275,842; 2,286,217; 42,649,692. So if we have 21 articles, the last article (Sundries) should get the 4 numbers before Total. That would be numbers 77-80. Then the total row is separate. So the 84 numbers include 20 articles 4 = 80 numbers for the first 20 articles, plus 4 for Sundries = 84? But 204=80, plus 4 =84. That would mean there are 20 articles before Sundries. But we have 21 articles before Sundries (if we count all except Building Materials). Let's count articles before Sundries in the OCR order (excluding Building Materials): Animals, Live; Chemicals and Drugs; Chinese Medicines; Dyeing & Tanning Materials; Foodstuffs & Provisions; Fuels; Hardware; Liquor, Intoxicating