§9.5.5

Row-Label Hierarchy

Indented row labels—how is the nesting carried?

✅ NormativeSettled. The requirement holds as written.📄 Page ruleCan be applied from the page alone.Read §9.5.5 in the RFC ↗

The requirement

Where a table's row labels are indented to show nesting, every label cell MUST carry its absolute level as a data-level attribute: <td data-level="2">iPhone</td>. Level 0 is the leftmost label position in the stub column, 1 the next indent step, and so on — counted from the left edge of the column, never from the row above.

When it applies

Financial statements and rate manuals encode hierarchy by indenting the row label. Flattened, "iPhone" and "Total net sales" read as siblings.

Sketch of the printed page — layout, not output

Printed stub:                     Rendered as:
Revenue                           <td data-level="0">Revenue</td>
  Products                        <td data-level="1">Products</td>
    iPhone          205,489       <td data-level="2">iPhone</td>
Total net sales     383,285       <th data-level="0">Total net sales</th>

A table whose indented labels all come out at the same level is the failure this rule exists to prevent.

Reading the examples. Each one pairs a source page with the output the rule expects. ✅ Sample: the rule applies. ❌ Counter-sample: it looks like it applies and does not, or a rendering got it wrong. A non-conforming output, in red, is shown only for contrast. … marks content left out of an excerpt.

Why flattening is dangerous

The flattened output is well-formed and looks correct. Nothing signals that a parent/child relationship was dropped, so a reader summing the column double-counts: iPhone is inside Total net sales, and the flattened table gives no way to know it. The damage is silent, which is why the indentation MUST be preserved even where the choice of encoding seems secondary.

❌ Counter-sampleLooks like it applies and does not — or a rendering got it wrong

Structure lost`3cc`

Under KVO shareholder the page indents two children, Renewable and Nuclear. The rendering below flattens them, leaving a second Renewable with no parent, and it also records 25 for KVO shareholder where the page prints 2.5. The total row is bold across all three cells.

Source page

Expected output · table 2 of 3

Energy production	Unit	2024
Renewable	TWh	29.5
Non-renewable	TWh	2.9
KVO shareholder	TWh	2.5
Renewable	TWh	0.4
Nuclear	TWh	2.1
Total energy production	TWh	34.8

Non-conforming output

What the rule rules out.

Energy production	Unit	2024
…
Non-renewable	TWh	2.9
KVO shareholder	TWh	25
Renewable	TWh	0.4
Nuclear	TWh	2.1
Total energy production	TWh	34.8

Reading the level

The level MUST be the number of indent steps between the left edge of the stub column and the start of the label as printed. The leftmost label position is 0; each further indent step adds one. A level can therefore be checked against a single row without reading any other.
The label text MUST stay verbatim and the rows MUST stay in printed order.
Where a table carries levels, every label cell in its stub that prints a label MUST carry one, level-0 rows included. A table whose stub shows no nesting — no indentation, no "of which" rows — and has no labelled total carries none.

Absolute, not relative

A level is read off the print, never inherited from the row above:

A row following a level-2 row is level 3 only if the page indents it further. At the same indent it is another level-2 sibling.
A row MAY jump back several levels at once. Total net sales returning to 0 after an iPhone at 2 is normal and needs no marker of its own.

Not from the numbers

A level MUST NOT be deepened because the arithmetic suggests a row is a component, and MUST NOT be flattened because the arithmetic does not add up. The indentation is the evidence; the values are not. A page whose sub-items do not sum to their parent is transcribed as printed.

Not as leading whitespace

The level MUST be carried as an attribute, not as leading spaces or  : the first trim, Markdown render or CSV round-trip removes whitespace and the structure goes with it. Where an output format cannot carry a cell attribute, the level SHOULD go into a dedicated level column rather than into whitespace.

Recognising a child row

A child is indented relative to the row above it, and its value is a component of a figure elsewhere in the column rather than an independent record. Phrases such as "out of which", "of which" and "within which" state the same relationship in words; a row introduced by one MUST be rendered as a child even where its indentation is slight. The same page may present a breakdown as an indented stub or as a row that decomposes into sub-rows; both are the same phenomenon and are rendered the same way.

A total row takes a level like any other row, usually 0; how its label cell is marked is defined in Total and Subtotal Rows (§9.5.6).

✅ SampleThe rule applies

"- Out of which amount written off" is indented under "Amount involved in frauds"`table-with-breakdown`

it is a component of that figure, not a peer, so it is level 1. The header labels wrap only for width.

Source page

Expected output

Provision debited to Profit and Loss Account	March 31, 2031	March 31, 2030
Number of frauds reported	9,742	13,208
Amount involved in frauds	184.30	121.55
- Out of which amount written off	0.00	0.62
Provisions made during the year (excluding write off)	41.20	5.90
Amount of unamortized provision debited from "other reserves" as at the end of the year	-	-

✅ SampleThe rule applies

Rule numbers nest into numbered and lettered sub-items`702`

609. Designated Business Pursuits › 3. Teachers › a. Lab, etc. / b. not otherwise classified / c. Corporal Punishment. The coverage labels sit at four printed indent positions, levels 0 to 3; the rule-number column is not part of the label hierarchy. Labels that wrap at the column edge are joined.

Source page

Expected output · excerpt

Basic Limit Rule#	Coverage F - Medical Payments	$2,000	$3,000	$4,000	$5,000
524.	Additional Insured - Household Resident	$ .58	$1.15	$1.73	$2.30
…
607.	Limited Home Day Care Coverage	2.88	5.75	8.63	10.93
608.	Permitted Incidental Occupancies - Residence Premises And Other Residences
	1. Residence Premises	2.88	5.75	8.63	10.93
	2. Other Residence	1.73	3.45	5.18	6.33
609.	Designated Business Pursuits
	1. Clerical Employees	.58	1.15	1.73	2.30
	2. Salesperson, Installation, etc. Included or Excluded	.58	1.15	1.73	2.30
	3. Teachers
	a. Lab, etc.	1.15	2.30	3.45	4.03
	b. not otherwise classified	.58	1.15	1.73	2.30
	c. Corporal Punishment	Medical Payments Not Available
611.	Incidental Low Power Recreational Motor Vehicles	.58	1.15	1.73	2.30
612.	Owned Watercraft Liability Coverage
	1. Outboard, Inboard or Inboard-Outboard engines or Motors
	a) Up to 15 feet:
	Up to 50 hp.	1.73	3.45	5.18	6.33
…

✅ SampleThe rule applies

Numbered parents (1)–(10) with indented lettered children`yellow`

Numbered parents (1)–(10) with indented lettered children — (A) Short / (B) Intermediate / (C) Long, (A) Dividends / (B) Capital gains — and an indented, unlettered dividends under (5) Preferred stock. The yellow fill marks input cells and carries no value, so it is not rendered.

Source page

Expected output · excerpt

	Invested Assets [1]	Currently Available Yield * [2]	Return On Invested Assets [3]=[1]*[2]	Federal Income Tax Rate [4]	Federal Income Taxes [5]=[3]*[4]
(1) US government bonds
(A) Short	34,614,395	0.13%	44,999	35.00%	15,750
(B) Intermediate	46,147,279	3.46%	1,596,696	35.00%	558,844
(C) Long	37,128,723	4.32%	1,603,961	35.00%	561,386
(2) Other taxable bonds
(A) Short	65,861,105	0.25%	164,653	35.00%	57,628
…
(4) Common Stock	22,262,218
(A) Dividends		4.63%	1,029,850	14.18%	145,981
(B) Capital gains		0.00%	0	34.10%	0
(5) Preferred stock	2,547,457
dividends		5.75%	146,479	14.18%	20,763
…

✅ SampleThe rule applies

Ticket and Transactions are indented under the Same Shop Sales parent rows`sssa`

The parent rows are bold in every cell, values included; the stub header Fiscal year ending December, is italic and not bold. Most percentages print a space before %, a few do not — kept as printed.

Source page

Expected output

	2023					2024
Fiscal year ending December,	Q1	Q2	Q3	Q4	FY	Q1	Q2	Q3
Systemwide Same Shop Sales	(2.0)%	3.8 %	4.0 %	5.0 %	2.8 %	10.0 %	4.1 %	2.7 %
Ticket	6.3 %	6.4 %	9.5 %	5.4 %	7.2 %	8.8 %	6.1 %	1.9 %
Transactions	(8.3)%	(2.6) %	(5.5) %	(0.4) %	(4.4) %	1.2 %	(2.0) %	0.8 %
Company-Operated Same Shop Sales	(3.5) %	1.6 %	2.8 %	4.6 %	1.5 %	10.9 %	5.2 %	4.0 %
Ticket	6.5 %	6.1 %	9.1 %	4.9 %	7.1 %	8.2 %	6.0 %	1.6 %
Transactions	(10.0) %	(4.5) %	(6.3) %	(0.3) %	(5.6) %	2.7 %	(0.8) %	2.4 %

✅ SampleThe rule applies

Bold TAMI / Professional Services parents over italic children (TAMI, Technology, …)`dt1`

The page prints no indentation: every label starts at the same left edge, so every row is level 0. The parent/child distinction is carried only by weight and slant, which apply to every cell of those rows and are rendered per cell. The table closes with a TOTAL row, so its stub carries levels.

Source page

Expected output · table 1 of 2, excerpt

	Firms	%	Square Footage Leased	%
TAMI	302	33%	10,444,741	42%
TAMI, Technology	120	13%	2,777,007	11%
TAMI, Advertising	52	6%	2,012,178	8%
TAMI, Marketing	13	1%	433,002	2%
TAMI, Media	98	11%	4,889,356	20%
TAMI, Public Relations	19	2%	208,331	1%
…

✅ SampleThe rule applies

細分 breakdown：批發 Wholesale and 其中 within which：航空運輸 Air transport open breakdowns of…`P200509160137_0137_6519`

The page prints no indentation; the relationship is stated in words, so each breakdown row is level 1, and the plain-weight rows that continue the same breakdown (零售 Retail, 陸上運輸 Land transport, 海運 Maritime transport) are level 1 with it.

Source page

Expected output · excerpt

	2005年 Q2指數 Index for Q2 2005	2005年Q2與 2005年Q1比較 Q2 2005 compared with Q1 2005		2005年Q2與 2004年Q2比較 Q2 2005 compared with Q2 2004		2005年Q1至Q2與 2004年Q1至Q2比較 Q1-Q2 2005 compared with Q1-Q2 2004
	（點） (Points)	（點） (Points)	（百分率） (%)	（點） (Points)	（百分率） (%)	（點） (Points)	（百分率） (%)
服務行業 Service Industry
批發/零售 Wholesale/Retail	102.3	-1.1	-1.1	+7.2	+7.6	+7.2	+7.5
細分 breakdown：批發 Wholesale	81.7	+1.3	+1.6	+6.6	+8.8	+4.8	+6.3
零售 Retail	109.7 *	-2.0	-1.8	+7.4	+7.3	+8.1	+7.9
進口與出口貿易 Import and export trades	108.5	+11.0	+11.3	+10.1	+10.3	+10.9	+11.8
飲食 Restaurants	92.5	-3.0	-3.1	+4.0	+4.5	+4.7	+5.3
酒店 Hotels	113.6	-0.9	-0.8	+20.8	+22.4	+21.0	+22.6
運輸 Transport	135.1	+9.9	+7.9	+17.7	+15.1	+18.9	+17.0
其中 within which：航空運輸 Air transport	139.6	+13.1	+10.4	+26.7	+23.6	+25.7	+23.9
陸上運輸 Land transport	107.3	+5.6	+5.5	+7.3	+7.3	+5.4	+5.5
海運 Maritime transport	159.4	+10.5	+7.1	+13.2	+9.0	+18.9	+13.9
…

Two more stub shapes

Two-column stubs

Where a table sets sub-labels in a second stub column to the right of the parent label, one per line, each with its own values, the output MUST render a two-column stub: the parent label in the first column, spanning its children with a rowspan where the page prints it as one cell, and each sub-label in the second column on its own row.

✅ SampleThe rule applies

a stub split into two columns

Unallocated interest charges sits in the first stub column, one ruled cell tall enough for both sub-labels, with – external and – intra-group in a second column to its right, each with its own values; the 6 months 2024 cell of – external is empty. Every group on the page closes with an unlabelled bold total. No label is indented, so every label — in either stub column — is level 0; the closing grand total is a labelled total, so its label cell is a <th>.

Source page

Expected output · excerpt

Continuing operations		6 months 2025 £m	6 months 2024 at 2025 exchange rates £m	6 months 2024 £m
Life achieved operating profit
United Kingdom		318	341	341
…
Other Europe		22	10	11
		279	224	227
Unallocated interest charges	– external	(74)	(58)
Unallocated interest charges	– intra-group	(76)	(91)	(91)
*Group operating profit before tax – continuing operations**		574	551	560

✅ SampleThe rule applies

values staggered inside one column

A Japanese profit-and-loss statement with one 金額 (amount) column: parent amounts right-aligned at the column edge, children set further left, and （うち…） ("of which") sub-items further left still. One value column. The stagger is styling; the hierarchy is carried only by the stub's data-level: the bold categories are 0, their items 1, and the （うち…） rows, introduced by "of which", 2.

Source page

Expected output

科　　目	金　　額
経常収益	742
資金運用収益	538
（うち貸出金利息）	（58）
（うち有価証券利息配当金）	（364）
役務取引等収益	91
その他業務収益	112
その他経常収益	1
経常費用	1,988
資金調達費用	177
（うち預金利息）	（177）
役務取引等費用	72
その他業務費用	248
営業経費	1,475
その他経常費用	16
経常損失	1,246
税引前当期損失	1,246
当期損失	1,246

✅ SampleThe rule applies

a two-column stub, ruled

Statutory turnover and Proportionate EBITDA sit in the first stub column; – NorthStar Wireless and – Other Americas in a second column beside each, one per line, each with values; an unlabelled bold subtotal closes each pair. Here the rules run across the first column between the two lines, so Proportionate EBITDA and (before exceptional items) are two cells, not one spanning cell.

Source page

Expected output

Financial highlights		Year ended 31 March
		2025	2024	Increase
		£m	£m	%
Statutory turnover	– NorthStar Wireless	–	–
	– Other Americas	14	10	40
		14	10
Proportionate EBITDA	– NorthStar Wireless	1,942	1,714	13
(before exceptional items)	– Other Americas	21	(48)
		1,963	1,666

Rule id tables-row-label-hierarchy · tables rule hierarchy encoding

When it applies

Why flattening is dangerous

Reading the level

Absolute, not relative

Not from the numbers

Not as leading whitespace

Recognising a child row

Two more stub shapes

Two-column stubs

Neighbouring rules, and how they differ

More