diff --git a/chronicle/source_package.py b/chronicle/source_package.py
index d2c76ef9..7e7888ce 100644
--- a/chronicle/source_package.py
+++ b/chronicle/source_package.py
@@ -358,6 +358,7 @@
     ),
     "soi-table-4-3": Path("irs_soi/table_4_3"),
     "soi-state-2022": Path("irs_soi/state_2022"),
+    "soi-state-2023": Path("irs_soi/state_2023"),
     "soi-county-2022": Path("irs_soi/county_2022"),
     "soi-congressional-district-2022": Path("irs_soi/congressional_district_2022"),
     "soi-historic-table-2": Path("irs_soi/historic_table_2"),
@@ -374,7 +375,11 @@
     "soi-ira-traditional-contributions-2022": Path(
         "irs_soi/ira_traditional_contributions_2022"
     ),
+    "soi-ira-traditional-contributions-2023": Path(
+        "irs_soi/ira_traditional_contributions_2023"
+    ),
     "soi-ira-roth-contributions-2022": Path("irs_soi/ira_roth_contributions_2022"),
+    "soi-ira-roth-contributions-2023": Path("irs_soi/ira_roth_contributions_2023"),
     "ssa-annual-statistical-supplement-2025": Path(
         "ssa/annual_statistical_supplement_2025"
     ),
diff --git a/chronicle/sources/cells.py b/chronicle/sources/cells.py
index f4fcfe99..a94119f3 100644
--- a/chronicle/sources/cells.py
+++ b/chronicle/sources/cells.py
@@ -18,6 +18,7 @@
 from zipfile import ZipFile
 
 import openpyxl
+from openpyxl.cell.cell import MergedCell
 import xlrd
 
 from chronicle.epoch import EMIT_EPOCH, HASH_DOMAINS, Epoch, hash_domain
@@ -172,8 +173,9 @@ def source_cells_from_xlsx(
         if sheets and sheet.title not in sheets:
             continue
         formula_sheet = formula_workbook[sheet.title]
-        for row_index in range(1, sheet.max_row + 1):
-            for column_index in range(1, sheet.max_column + 1):
+        max_row, max_column = _xlsx_used_range_bounds(sheet)
+        for row_index in range(1, max_row + 1):
+            for column_index in range(1, max_column + 1):
                 cell = sheet.cell(row=row_index, column=column_index)
                 formula_cell = formula_sheet.cell(
                     row=row_index,
@@ -199,6 +201,48 @@ def source_cells_from_xlsx(
     return cells
 
 
+# The last column (XFD) and row an .xlsx worksheet can address.
+_XLSX_MAX_COLUMN = 16_384
+_XLSX_MAX_ROW = 1_048_576
+
+
+def _xlsx_used_range_bounds(sheet: Any) -> tuple[int, int]:
+    """Return the (max_row, max_column) of a worksheet's used range.
+
+    This is openpyxl's ``(sheet.max_row, sheet.max_column)`` unless the sheet
+    has a merged range that runs to the worksheet edge. openpyxl fills every
+    position of a merged range with a ``MergedCell`` placeholder, so one
+    footnote row merged across the whole sheet width (``A179:XFD179`` in the
+    IRS's TY2023 ``23in54us.xlsx``) stretches ``max_column`` to 16,384 and the
+    used range to about 70 million empty cells. Such an edge-to-edge range is
+    sheet-width formatting, not table extent: its placeholders do not widen
+    the bounds, while its anchor cell and every other cell still do. Sheets
+    without one keep openpyxl's bounds exactly.
+    """
+    edge_ranges = [
+        cell_range
+        for cell_range in sheet.merged_cells.ranges
+        if cell_range.max_col >= _XLSX_MAX_COLUMN or cell_range.max_row >= _XLSX_MAX_ROW
+    ]
+    if not edge_ranges:
+        return sheet.max_row, sheet.max_column
+
+    def edge_placeholder(row_index: int, column_index: int, cell: Any) -> bool:
+        return isinstance(cell, MergedCell) and any(
+            cell_range.min_row <= row_index <= cell_range.max_row
+            and cell_range.min_col <= column_index <= cell_range.max_col
+            for cell_range in edge_ranges
+        )
+
+    max_row = max_column = 1
+    for (row_index, column_index), cell in sheet._cells.items():
+        if edge_placeholder(row_index, column_index, cell):
+            continue
+        max_row = max(max_row, row_index)
+        max_column = max(max_column, column_index)
+    return max_row, max_column
+
+
 def source_cells_from_ods(
     content: bytes,
     artifact: SourceArtifactMetadata,
diff --git a/chronicle/targets/us_poverty.py b/chronicle/targets/us_poverty.py
index cfc5f3af..607134d2 100644
--- a/chronicle/targets/us_poverty.py
+++ b/chronicle/targets/us_poverty.py
@@ -85,6 +85,7 @@ def has_chronicle_package(self) -> bool:
             "soi-filing-season-week47-2024-eitc-total",
             "soi-table-4-3",
             "soi-state-2022",
+            "soi-state-2023",
             "soi-historic-table-2",
             "soi-historic-table-2-state-agi-2022",
             "soi-historic-table-2-state-broad-2022",
diff --git a/db/data/irs_soi/ira_contributions/23in05ira.xlsx b/db/data/irs_soi/ira_contributions/23in05ira.xlsx
new file mode 100644
index 00000000..982e0870
Binary files /dev/null and b/db/data/irs_soi/ira_contributions/23in05ira.xlsx differ
diff --git a/db/data/irs_soi/ira_contributions/23in06ira.xlsx b/db/data/irs_soi/ira_contributions/23in06ira.xlsx
new file mode 100644
index 00000000..da122a3a
Binary files /dev/null and b/db/data/irs_soi/ira_contributions/23in06ira.xlsx differ
diff --git a/db/data/irs_soi/ira_contributions/manifest_roth_2023_source_package.yaml b/db/data/irs_soi/ira_contributions/manifest_roth_2023_source_package.yaml
new file mode 100644
index 00000000..ff9d3f98
--- /dev/null
+++ b/db/data/irs_soi/ira_contributions/manifest_roth_2023_source_package.yaml
@@ -0,0 +1,21 @@
+source_id: irs_soi
+package_id: soi-ira-roth-contributions-2023
+dataset: irs_soi_roth_ira_contributions_2023
+source_page: https://www.irs.gov/statistics/soi-tax-stats-accumulation-and-distribution-of-individual-retirement-arrangements
+table: Table 6. Taxpayers with Roth Individual Retirement Arrangement (IRA) Plan Contributions,
+  by Size of Contribution and Age of Taxpayer
+files:
+  2023:
+    filename: 23in06ira.xlsx
+    source_url: https://www.irs.gov/pub/irs-soi/23in06ira.xlsx
+    sha256: b526e4d4163172ed9992d16375cbf601b1c8b4d5bcf66f86e038c0e91c420418
+    size_bytes: 11380
+    fetched_at: '2026-09-27T05:55:43+00:00'
+    source_table: Table 6. Taxpayers with Roth Individual Retirement Arrangement (IRA)
+      Plan Contributions, by Size of Contribution and Age of Taxpayer
+    storage:
+      r2:
+        provider: r2
+        bucket: ledger-raw
+        key: raw/irs_soi/soi-ira-roth-contributions-2023/2023/b526e4d4163172ed9992d16375cbf601b1c8b4d5bcf66f86e038c0e91c420418/23in06ira.xlsx
+        uri: r2://ledger-raw/raw/irs_soi/soi-ira-roth-contributions-2023/2023/b526e4d4163172ed9992d16375cbf601b1c8b4d5bcf66f86e038c0e91c420418/23in06ira.xlsx
diff --git a/db/data/irs_soi/ira_contributions/manifest_traditional_2023_source_package.yaml b/db/data/irs_soi/ira_contributions/manifest_traditional_2023_source_package.yaml
new file mode 100644
index 00000000..62d676e9
--- /dev/null
+++ b/db/data/irs_soi/ira_contributions/manifest_traditional_2023_source_package.yaml
@@ -0,0 +1,21 @@
+source_id: irs_soi
+package_id: soi-ira-traditional-contributions-2023
+dataset: irs_soi_traditional_ira_contributions_2023
+source_page: https://www.irs.gov/statistics/soi-tax-stats-accumulation-and-distribution-of-individual-retirement-arrangements
+table: Table 5. Taxpayers with Traditional Individual Retirement Arrangement (IRA)
+  Plan Contributions, by Size of Contribution and Age of Taxpayer
+files:
+  2023:
+    filename: 23in05ira.xlsx
+    source_url: https://www.irs.gov/pub/irs-soi/23in05ira.xlsx
+    sha256: 17b02670eeb35dab5d3f8be6a26333d527993f8b22933fa77bb3a06f3c8ad143
+    size_bytes: 11345
+    fetched_at: '2026-09-27T05:55:45+00:00'
+    source_table: Table 5. Taxpayers with Traditional Individual Retirement Arrangement
+      (IRA) Plan Contributions, by Size of Contribution and Age of Taxpayer
+    storage:
+      r2:
+        provider: r2
+        bucket: ledger-raw
+        key: raw/irs_soi/soi-ira-traditional-contributions-2023/2023/17b02670eeb35dab5d3f8be6a26333d527993f8b22933fa77bb3a06f3c8ad143/23in05ira.xlsx
+        uri: r2://ledger-raw/raw/irs_soi/soi-ira-traditional-contributions-2023/2023/17b02670eeb35dab5d3f8be6a26333d527993f8b22933fa77bb3a06f3c8ad143/23in05ira.xlsx
diff --git a/db/data/irs_soi/state_2023/23in54us.xlsx b/db/data/irs_soi/state_2023/23in54us.xlsx
new file mode 100644
index 00000000..de69dddd
Binary files /dev/null and b/db/data/irs_soi/state_2023/23in54us.xlsx differ
diff --git a/db/data/irs_soi/state_2023/manifest.yaml b/db/data/irs_soi/state_2023/manifest.yaml
new file mode 100644
index 00000000..de87c4cc
--- /dev/null
+++ b/db/data/irs_soi/state_2023/manifest.yaml
@@ -0,0 +1,18 @@
+source_id: irs_soi
+package_id: soi-state-2023
+dataset: irs_soi_state_2023
+source_page: https://www.irs.gov/statistics/soi-tax-stats-historic-table-2
+table: IRS SOI State Data 2023
+files:
+  2023:
+    filename: 23in54us.xlsx
+    source_url: https://www.irs.gov/pub/irs-soi/23in54us.xlsx
+    sha256: 4bc45f2586fb924c1e0e7818773096bdcf71223b6407120ca058efaed728607a
+    size_bytes: 265549
+    fetched_at: '2026-09-27T05:55:34+00:00'
+    storage:
+      r2:
+        provider: r2
+        bucket: ledger-raw
+        key: raw/irs_soi/soi-state-2023/2023/4bc45f2586fb924c1e0e7818773096bdcf71223b6407120ca058efaed728607a/23in54us.xlsx
+        uri: r2://ledger-raw/raw/irs_soi/soi-state-2023/2023/4bc45f2586fb924c1e0e7818773096bdcf71223b6407120ca058efaed728607a/23in54us.xlsx
diff --git a/docs/pe-calibration-targets.md b/docs/pe-calibration-targets.md
index 899a8c39..82fbb12b 100644
--- a/docs/pe-calibration-targets.md
+++ b/docs/pe-calibration-targets.md
@@ -36,7 +36,7 @@ them to model variables.
 |---|---|---|
 | Population by age, sex, state, and congressional district | Census PEP and ACS demographics | `census-pep-2024-national-age-sex`, `census-pep-2024-state-age-sex`, `census-acs-s0101-national-age-2024`, `census-acs-s0101-state-age-2024`, `census-acs-s0101-congressional-district-age-2024` |
 | NIPA personal income, transfers, taxes, and pensions | BEA NIPA full-population aggregates | `bea-nipa-total-wages-salaries`, `bea-nipa-personal-income-components`, `bea-nipa-personal-income-disposition`, `bea-nipa-pension-contributions` |
-| SOI filer income, taxes, deductions, and credits | IRS SOI administrative totals | `soi-table-1-1`, `soi-table-1-2`, `soi-table-1-4`, `soi-table-2-1`, `soi-table-2-5`, `soi-table-2-5-eitc-agi-children-2023`, `soi-table-4-3`, `soi-state-2022`, `soi-historic-table-2`, `soi-historic-table-2-state-agi-2022`, `soi-historic-table-2-state-broad-2022`, `soi-historic-table-2-state-eitc-2022`, `soi-w2-statistics-2020` |
+| SOI filer income, taxes, deductions, and credits | IRS SOI administrative totals | `soi-table-1-1`, `soi-table-1-2`, `soi-table-1-4`, `soi-table-2-1`, `soi-table-2-5`, `soi-table-2-5-eitc-agi-children-2023`, `soi-table-4-3`, `soi-state-2022`, `soi-state-2023`, `soi-historic-table-2`, `soi-historic-table-2-state-agi-2022`, `soi-historic-table-2-state-broad-2022`, `soi-historic-table-2-state-eitc-2022`, `soi-w2-statistics-2020` |
 | Social Security and SSI | SSA administrative totals | `ssa-annual-statistical-supplement-2025`, `ssa-ssi-table-7b1-2024` |
 | SNAP | USDA FNS administrative totals | `usda-snap-fy69-to-current` |
 | TANF | HHS ACF caseload and financial totals | `hhs-acf-tanf-caseload-2024`, `hhs-acf-tanf-financial-2024` |
diff --git a/packages/irs_soi/ira_roth_contributions_2023/source_package.yaml b/packages/irs_soi/ira_roth_contributions_2023/source_package.yaml
new file mode 100644
index 00000000..8ce32df9
--- /dev/null
+++ b/packages/irs_soi/ira_roth_contributions_2023/source_package.yaml
@@ -0,0 +1,69 @@
+schema_version: ledger.source_package.v1
+package_id: soi-ira-roth-contributions-2023
+label: IRS SOI 2023 Roth IRA contribution totals
+dimension_labels:
+  'irs_soi.ira_contribution_size': 'IRA contribution size'
+artifact:
+  source_name: irs_soi
+  source_table: Table 6. Taxpayers with Roth Individual Retirement Arrangement (IRA) Plan Contributions, by Size of Contribution and Age of Taxpayer
+  resource_package: db
+  resource_directory: data/irs_soi/ira_contributions
+  manifest: manifest_roth_2023_source_package.yaml
+  vintage: tax_year_2023
+  extracted_at: '2026-09-27'
+  extraction_method: xlsx whole-workbook used-range cell parse
+  parser: xlsx_used_range
+  artifact_year: 2023
+record_sets:
+- record_set_id: irs_soi.ty2023.roth_ira_contributions.all_taxpayers
+  provenance_class: administrative
+  record_set_spec_id: irs_soi.roth_ira_contributions.all_taxpayers.v1
+  source_record_id_prefix: irs_soi.ty2023.roth_ira_contributions.all_taxpayers
+  sheet_name: Estimates
+  period_type: tax_year
+  period: '2023'
+  geography_id: 0100000US
+  geography_level: country
+  geography_name: United States
+  geography_vintage: current
+  entity: tax_unit
+  entity_role: filing_unit
+  domain: individual_retirement_arrangement_contributions
+  groupby_dimension: irs_soi.ira_contribution_size
+  rows:
+  - value_id: all_taxpayers
+    label: All taxpayers
+    ordinal: 0
+    row_number: 7
+    expected_row_header_column: A
+    expected_row_header: All taxpayers
+    table_record_kind: total
+    guard_cells:
+    - column: B
+      row: 4
+      expected_value: Total
+      label: total contribution size
+  measures:
+  - measure_id: taxpayer_count
+    label: Number of taxpayers
+    ordinal: 0
+    column: B
+    source_column_id: number_of_taxpayers
+    expected_column_header_row: 5
+    expected_column_header: Number of taxpayers
+    concept: irs_soi.roth_ira_contributors
+    unit: count
+    aggregation: sum
+    expected_cell_type: number
+  - measure_id: amount
+    label: Contribution amount
+    ordinal: 1
+    column: C
+    source_column_id: amount
+    expected_column_header_row: 5
+    expected_column_header: Amount
+    concept: irs_soi.roth_ira_contributions
+    unit: usd
+    aggregation: sum
+    value_scale: 1000
+    expected_cell_type: number
diff --git a/packages/irs_soi/ira_traditional_contributions_2023/source_package.yaml b/packages/irs_soi/ira_traditional_contributions_2023/source_package.yaml
new file mode 100644
index 00000000..e1b3ce50
--- /dev/null
+++ b/packages/irs_soi/ira_traditional_contributions_2023/source_package.yaml
@@ -0,0 +1,69 @@
+schema_version: ledger.source_package.v1
+package_id: soi-ira-traditional-contributions-2023
+label: IRS SOI 2023 Traditional IRA contribution totals
+dimension_labels:
+  'irs_soi.ira_contribution_size': 'IRA contribution size'
+artifact:
+  source_name: irs_soi
+  source_table: Table 5. Taxpayers with Traditional Individual Retirement Arrangement (IRA) Plan Contributions, by Size of Contribution and Age of Taxpayer
+  resource_package: db
+  resource_directory: data/irs_soi/ira_contributions
+  manifest: manifest_traditional_2023_source_package.yaml
+  vintage: tax_year_2023
+  extracted_at: '2026-09-27'
+  extraction_method: xlsx whole-workbook used-range cell parse
+  parser: xlsx_used_range
+  artifact_year: 2023
+record_sets:
+- record_set_id: irs_soi.ty2023.traditional_ira_contributions.all_taxpayers
+  provenance_class: administrative
+  record_set_spec_id: irs_soi.traditional_ira_contributions.all_taxpayers.v1
+  source_record_id_prefix: irs_soi.ty2023.traditional_ira_contributions.all_taxpayers
+  sheet_name: Estimates
+  period_type: tax_year
+  period: '2023'
+  geography_id: 0100000US
+  geography_level: country
+  geography_name: United States
+  geography_vintage: current
+  entity: tax_unit
+  entity_role: filing_unit
+  domain: individual_retirement_arrangement_contributions
+  groupby_dimension: irs_soi.ira_contribution_size
+  rows:
+  - value_id: all_taxpayers
+    label: All taxpayers
+    ordinal: 0
+    row_number: 7
+    expected_row_header_column: A
+    expected_row_header: All taxpayers
+    table_record_kind: total
+    guard_cells:
+    - column: B
+      row: 4
+      expected_value: Total
+      label: total contribution size
+  measures:
+  - measure_id: taxpayer_count
+    label: Number of taxpayers
+    ordinal: 0
+    column: B
+    source_column_id: number_of_taxpayers
+    expected_column_header_row: 5
+    expected_column_header: Number of taxpayers
+    concept: irs_soi.traditional_ira_contributors
+    unit: count
+    aggregation: sum
+    expected_cell_type: number
+  - measure_id: amount
+    label: Contribution amount
+    ordinal: 1
+    column: C
+    source_column_id: amount
+    expected_column_header_row: 5
+    expected_column_header: Amount
+    concept: irs_soi.traditional_ira_contributions
+    unit: usd
+    aggregation: sum
+    value_scale: 1000
+    expected_cell_type: number
diff --git a/packages/irs_soi/state_2023/source_package.yaml b/packages/irs_soi/state_2023/source_package.yaml
new file mode 100644
index 00000000..3794862a
--- /dev/null
+++ b/packages/irs_soi/state_2023/source_package.yaml
@@ -0,0 +1,228 @@
+schema_version: ledger.source_package.v1
+package_id: soi-state-2023
+label: IRS SOI 2023 state-data United States totals
+dimension_labels:
+  'filing_status': 'Filing status'
+  'income_range': 'Income range'
+  'irs_soi.state_table_measure': 'State table measure'
+  'us.tax.earned_income_credit_qualifying_children': 'EITC qualifying children'
+dimension_value_labels:
+  'filing_status':
+    'all': 'All filing statuses'
+  'income_range':
+    'all': 'All income ranges'
+artifact:
+  source_name: irs_soi
+  source_table: Historic Table 2 state data, United States total
+  resource_package: db
+  resource_directory: data/irs_soi/state_2023
+  manifest: manifest.yaml
+  vintage: tax_year_2023
+  extracted_at: "2026-09-27"
+  extraction_method: xlsx whole-workbook used-range cell parse
+  parser: xlsx_used_range
+  artifact_year: 2023
+record_sets:
+  - record_set_id: irs_soi.ty2023.state_2023.us.return_count
+    provenance_class: administrative
+    record_set_spec_id: irs_soi.state_2022.us.return_count.v1
+    source_record_id_prefix: irs_soi.ty2023.state_2023.us.return_count
+    sheet_name: Sheet1
+    period_type: tax_year
+    period: "2023"
+    geography_id: 0100000US
+    geography_level: country
+    geography_name: United States
+    geography_vintage: 2020_census
+    entity: tax_unit
+    entity_role: filing_unit
+    domain: all_individual_income_tax_returns
+    groupby_dimension: irs_soi.state_table_measure
+    rows:
+      - value_id: all_returns
+        label: All returns
+        ordinal: 0
+        row_number: 9
+        expected_row_header_column: A
+        expected_row_header: Number of returns [1]
+        filters:
+          filing_status: all
+          income_range: all
+        guard_cells:
+          - column: B
+            row: 3
+            expected_value: "UNITED STATES "
+            label: geography section
+        table_record_kind: total
+    measures:
+      - measure_id: return_count
+        label: Number of returns
+        ordinal: 0
+        column: B
+        source_column_id: all_returns
+        expected_column_header_row: 4
+        expected_column_header: All returns
+        concept: irs_soi.individual_income_tax_returns
+        unit: count
+        aggregation: sum
+        expected_cell_type: number
+  - record_set_id: irs_soi.ty2023.state_2023.us.adjusted_gross_income
+    provenance_class: administrative
+    record_set_spec_id: irs_soi.state_2022.us.adjusted_gross_income.v1
+    source_record_id_prefix: irs_soi.ty2023.state_2023.us.adjusted_gross_income
+    sheet_name: Sheet1
+    period_type: tax_year
+    period: "2023"
+    geography_id: 0100000US
+    geography_level: country
+    geography_name: United States
+    geography_vintage: 2020_census
+    entity: tax_unit
+    entity_role: filing_unit
+    domain: all_individual_income_tax_returns
+    groupby_dimension: irs_soi.state_table_measure
+    rows:
+      - value_id: all_returns
+        label: All returns
+        ordinal: 0
+        row_number: 26
+        expected_row_header_column: A
+        expected_row_header: Adjusted gross income (AGI) [6]
+        filters:
+          filing_status: all
+          income_range: all
+        guard_cells:
+          - column: B
+            row: 3
+            expected_value: "UNITED STATES "
+            label: geography section
+        table_record_kind: total
+    measures:
+      - measure_id: amount
+        label: Adjusted gross income
+        ordinal: 0
+        column: B
+        source_column_id: all_returns
+        expected_column_header_row: 4
+        expected_column_header: All returns
+        concept: us:statutes/26/62#adjusted_gross_income
+        source_concept: irs_soi.adjusted_gross_income
+        concept_relation: exact
+        concept_authority: ledger-us
+        concept_evidence_url: https://uscode.house.gov/view.xhtml?req=(title:26%20section:62%20edition:prelim)
+        concept_evidence_notes: >
+          IRS SOI Historic Table 2 state-data files report adjusted gross
+          income for individual income tax returns; IRC section 62 defines
+          adjusted gross income. This Ledger assertion treats the SOI AGI row
+          as exactly adopting that legal concept for the tax-year source
+          record.
+        legal_vintage: tax_year_2023
+        unit: usd
+        aggregation: sum
+        value_scale: 1000
+        expected_cell_type: number
+  - record_set_id: irs_soi.ty2023.state_2023.us.eitc_three_or_more_children_returns
+    provenance_class: administrative
+    record_set_spec_id: irs_soi.state_2022.us.eitc_three_or_more_children_returns.v1
+    source_record_id_prefix: irs_soi.ty2023.state_2023.us.eitc_three_or_more_children_returns
+    sheet_name: Sheet1
+    period_type: tax_year
+    period: "2023"
+    geography_id: 0100000US
+    geography_level: country
+    geography_name: United States
+    geography_vintage: 2020_census
+    entity: tax_unit
+    entity_role: filing_unit
+    domain: all_individual_income_tax_returns
+    groupby_dimension: us.tax.earned_income_credit_qualifying_children
+    rows:
+      - value_id: three_or_more_qualifying_children
+        label: Three or more qualifying children
+        ordinal: 0
+        row_number: 146
+        expected_row_header_column: A
+        expected_row_header: "         Earned income credit with three or more qualifying children:  Number"
+        filters:
+          filing_status: all
+          income_range: all
+        constraints:
+          - variable: us.tax.earned_income_credit_qualifying_children
+            operator: ">="
+            value: 3
+            unit: count
+            label: Earned income credit qualifying children lower bound
+        guard_cells:
+          - column: B
+            row: 3
+            expected_value: "UNITED STATES "
+            label: geography section
+          - column: A
+            row: 138
+            expected_value: "Earned income credit:  [10] Number"
+            label: earned income credit section
+    measures:
+      - measure_id: return_count
+        label: Returns with earned income credit
+        ordinal: 0
+        column: B
+        source_column_id: all_returns
+        expected_column_header_row: 4
+        expected_column_header: All returns
+        concept: irs_soi.returns_with_earned_income_credit
+        unit: count
+        aggregation: sum
+        expected_cell_type: number
+  - record_set_id: irs_soi.ty2023.state_2023.us.eitc_three_or_more_children_amount
+    provenance_class: administrative
+    record_set_spec_id: irs_soi.state_2022.us.eitc_three_or_more_children_amount.v1
+    source_record_id_prefix: irs_soi.ty2023.state_2023.us.eitc_three_or_more_children_amount
+    sheet_name: Sheet1
+    period_type: tax_year
+    period: "2023"
+    geography_id: 0100000US
+    geography_level: country
+    geography_name: United States
+    geography_vintage: 2020_census
+    entity: tax_unit
+    entity_role: filing_unit
+    domain: all_individual_income_tax_returns
+    groupby_dimension: us.tax.earned_income_credit_qualifying_children
+    rows:
+      - value_id: three_or_more_qualifying_children
+        label: Three or more qualifying children
+        ordinal: 0
+        row_number: 147
+        expected_row_header_column: A
+        expected_row_header: "     Amount"
+        filters:
+          filing_status: all
+          income_range: all
+        constraints:
+          - variable: us.tax.earned_income_credit_qualifying_children
+            operator: ">="
+            value: 3
+            unit: count
+            label: Earned income credit qualifying children lower bound
+        guard_cells:
+          - column: B
+            row: 3
+            expected_value: "UNITED STATES "
+            label: geography section
+          - column: A
+            row: 146
+            expected_value: "         Earned income credit with three or more qualifying children:  Number"
+            label: earned income credit qualifying children row
+    measures:
+      - measure_id: amount
+        label: Earned income credit
+        ordinal: 0
+        column: B
+        source_column_id: all_returns
+        expected_column_header_row: 4
+        expected_column_header: All returns
+        concept: irs_soi.earned_income_credit
+        unit: usd
+        aggregation: sum
+        value_scale: 1000
+        expected_cell_type: number
diff --git a/tests/test_chronicle_bundle.py b/tests/test_chronicle_bundle.py
index 4d6fcd36..a8866dd6 100644
--- a/tests/test_chronicle_bundle.py
+++ b/tests/test_chronicle_bundle.py
@@ -136,13 +136,13 @@ def test_build_bundle_writes_merged_consumer_contract(tmp_path):
         "aggregate_duplicate_key_count": 0,
         "entity_count": 12,
         "error_count": 0,
-        "fact_count": 350893,
+        "fact_count": 350901,
         "geography_count": 12592,
         "period_count": 494,
         "semantic_duplicate_key_count": 467,
         "skipped_source_count": 10,
         "source_count": 50,
-        "source_package_count": 227,
+        "source_package_count": 230,
         # 1 semantic-duplicate warning, plus the publisher wording Chronicle
         # keeps as published: values two packages word differently, groupby
         # rows that drift inside one package (chronicle#265, #266), and the
@@ -150,7 +150,7 @@ def test_build_bundle_writes_merged_consumer_contract(tmp_path):
         # it does has stopped warning (chronicle#281).
         "warning_count": 76,
     }
-    assert len(rows) == 350893
+    assert len(rows) == 350901
     assert {row["provenance_class"] for row in rows} <= {
         "administrative",
         "census",
@@ -168,7 +168,7 @@ def test_build_bundle_writes_merged_consumer_contract(tmp_path):
     )
     assert rows[0]["aggregate_fact_key"].startswith("ledger.aggregate_fact.v2:")
     assert rows[0]["semantic_fact_key"].startswith("ledger.semantic_fact.v2:")
-    assert source_packages["source_package_count"] == 227
+    assert source_packages["source_package_count"] == 230
     assert source_packages["skipped_source_count"] == 10
     assert sorted(item["source"] for item in source_packages["skipped_sources"]) == [
         "census-acs-s0101-congressional-district-age-2024",
@@ -182,7 +182,7 @@ def test_build_bundle_writes_merged_consumer_contract(tmp_path):
         "jct-obbba-revenue-estimates-2025",
         "jct-tax-expenditures-2024",
     ]
-    assert coverage["fact_count"] == 350893
+    assert coverage["fact_count"] == 350901
     assert coverage["counts"]["by_source"] == {
         "bea": 445,
         "bfp_economic_outlook": 5,
@@ -207,7 +207,7 @@ def test_build_bundle_writes_merged_consumer_contract(tmp_path):
         "hhs_acf_tanf": 110,
         "hmrc": 31343,
         "ici": 12,
-        "irs_soi": 40063,
+        "irs_soi": 40071,
         "isc": 2,
         "jrc_euromod_be": 90,
         "kff": 52,
@@ -1190,6 +1190,8 @@ def test_build_bundle_writes_merged_consumer_contract(tmp_path):
         issue_280_period_increments[f"month:{month}"] = 23
     for key, count in issue_280_period_increments.items():
         expected_period_counts[key] = expected_period_counts.get(key, 0) + count
+    # IRS SOI TY2023 state-data US totals (4) and IRA Tables 5/6 (2 + 2).
+    expected_period_counts["tax_year:2023"] += 8
     assert coverage["counts"]["by_period"] == expected_period_counts
     assert coverage["counts"]["by_geography"]["country:BE"] == 4888
     assert coverage["counts"]["by_geography"]["country:DE"] == 36
@@ -1198,7 +1200,7 @@ def test_build_bundle_writes_merged_consumer_contract(tmp_path):
     assert coverage["counts"]["by_geography"]["nuts1:BE2"] == 3673
     assert coverage["counts"]["by_geography"]["nuts1:BE3"] == 3662
     assert coverage["counts"]["by_geography"]["commune:11002"] == 1
-    assert coverage["counts"]["by_geography"]["country:0100000US"] == 2109
+    assert coverage["counts"]["by_geography"]["country:0100000US"] == 2117
     assert coverage["counts"]["by_geography"]["state:0400000US06"] == 229
     assert (
         coverage["counts"]["by_geography"]["congressional_district:5001700US0601"] == 56
@@ -1220,7 +1222,7 @@ def test_build_bundle_writes_merged_consumer_contract(tmp_path):
         "person": 74808,
         "return": 14600,
         "social_protection_scheme": 36,
-        "tax_unit": 41368,
+        "tax_unit": 41376,
     }
     assert not coverage["duplicates"]["aggregate_fact_keys"]
     assert len(coverage["duplicates"]["semantic_fact_keys"]) == 467
diff --git a/tests/test_chronicle_soi_state_ira_2023.py b/tests/test_chronicle_soi_state_ira_2023.py
new file mode 100644
index 00000000..f57d876b
--- /dev/null
+++ b/tests/test_chronicle_soi_state_ira_2023.py
@@ -0,0 +1,269 @@
+"""IRS SOI TY2023 state-data US totals and IRA Tables 5/6.
+
+``soi-state-2023`` reads the United States workbook of the TY2023 state data
+(``23in54us.xlsx``); ``soi-ira-traditional-contributions-2023`` and
+``soi-ira-roth-contributions-2023`` read IRA Tables 5 and 6 of the June 2026
+IRA study (``23in05ira.xlsx``, ``23in06ira.xlsx``). Each mirrors its TY2022
+twin with literal TY2023 labels. The workbooks' layouts moved, and the
+packages move with them:
+
+- 23in54us.xlsx: the geography label moved from A8 to B3, the "All returns"
+  header from row 3 to row 4, and the EITC block two rows down.
+- The IRA tables use an "Estimates" sheet with "All taxpayers" on row 7.
+
+The expected values are read with openpyxl directly, not through Chronicle's
+cell parser.
+"""
+
+from __future__ import annotations
+
+import csv
+import hashlib
+import json
+from pathlib import Path
+
+import openpyxl
+import pytest
+import yaml
+
+from chronicle.core import validate_facts
+from chronicle.source_package import SOURCE_PACKAGE_ALIASES, load_source_package
+from chronicle.sources.cells import validate_source_cells
+from chronicle.suite import build_source_suite
+
+REPO_ROOT = Path(__file__).resolve().parents[1]
+DATA = REPO_ROOT / "db" / "data" / "irs_soi"
+PACKAGES_ROOT = REPO_ROOT / "packages"
+
+# package -> (TY2022 twin, manifest, workbook, sha256, size, sheet,
+#             {source_record_id suffix: (cell, scale)})
+PACKAGES = {
+    "soi-state-2023": (
+        "soi-state-2022",
+        DATA / "state_2023" / "manifest.yaml",
+        DATA / "state_2023" / "23in54us.xlsx",
+        "4bc45f2586fb924c1e0e7818773096bdcf71223b6407120ca058efaed728607a",
+        265_549,
+        "Sheet1",
+        {
+            "state_2023.us.return_count.all_returns.return_count": ("B9", 1),
+            "state_2023.us.adjusted_gross_income.all_returns.amount": ("B26", 1000),
+            (
+                "state_2023.us.eitc_three_or_more_children_returns."
+                "three_or_more_qualifying_children.return_count"
+            ): ("B146", 1),
+            (
+                "state_2023.us.eitc_three_or_more_children_amount."
+                "three_or_more_qualifying_children.amount"
+            ): ("B147", 1000),
+        },
+    ),
+    "soi-ira-traditional-contributions-2023": (
+        "soi-ira-traditional-contributions-2022",
+        DATA / "ira_contributions" / "manifest_traditional_2023_source_package.yaml",
+        DATA / "ira_contributions" / "23in05ira.xlsx",
+        "17b02670eeb35dab5d3f8be6a26333d527993f8b22933fa77bb3a06f3c8ad143",
+        11_345,
+        "Estimates",
+        {
+            "traditional_ira_contributions.all_taxpayers.all_taxpayers.taxpayer_count": (
+                "B7",
+                1,
+            ),
+            "traditional_ira_contributions.all_taxpayers.all_taxpayers.amount": (
+                "C7",
+                1000,
+            ),
+        },
+    ),
+    "soi-ira-roth-contributions-2023": (
+        "soi-ira-roth-contributions-2022",
+        DATA / "ira_contributions" / "manifest_roth_2023_source_package.yaml",
+        DATA / "ira_contributions" / "23in06ira.xlsx",
+        "b526e4d4163172ed9992d16375cbf601b1c8b4d5bcf66f86e038c0e91c420418",
+        11_380,
+        "Estimates",
+        {
+            "roth_ira_contributions.all_taxpayers.all_taxpayers.taxpayer_count": (
+                "B7",
+                1,
+            ),
+            "roth_ira_contributions.all_taxpayers.all_taxpayers.amount": ("C7", 1000),
+        },
+    ),
+}
+PUBLISHED = {
+    # Spot values from the publisher files, as printed.
+    "irs_soi.ty2023.state_2023.us.return_count.all_returns.return_count": 159_949_000,
+    "irs_soi.ty2023.traditional_ira_contributions.all_taxpayers.all_taxpayers."
+    "taxpayer_count": 5_873_053,
+    "irs_soi.ty2023.roth_ira_contributions.all_taxpayers.all_taxpayers."
+    "taxpayer_count": 9_488_414,
+}
+
+
+def _build(package_id: str, year: int):
+    package = load_source_package(package_id)
+    rows = package.build_source_rows(year)
+    cells = package.build_source_cells(year, source_rows=rows)
+    return cells, package.build_facts(year, cells=cells, source_rows=rows)
+
+
+@pytest.fixture(scope="module")
+def built():
+    facts_by_package = {}
+    for package_id in PACKAGES:
+        cells, facts = _build(package_id, 2023)
+        assert validate_source_cells(cells).valid
+        assert validate_facts(facts).valid
+        facts_by_package[package_id] = facts
+    return facts_by_package
+
+
+def _cell(workbook: Path, sheet: str, address: str):
+    book = openpyxl.load_workbook(workbook, read_only=True, data_only=True)
+    try:
+        return book[sheet][address].value
+    finally:
+        book.close()
+
+
+@pytest.mark.parametrize("package_id", sorted(PACKAGES))
+def test_manifest_registers_the_publisher_file(package_id):
+    _, manifest_path, workbook, sha256, size, _, _ = PACKAGES[package_id]
+    entry = yaml.safe_load(manifest_path.read_text())["files"][2023]
+
+    assert hashlib.sha256(workbook.read_bytes()).hexdigest() == sha256
+    assert entry["filename"] == workbook.name
+    assert entry["source_url"] == f"https://www.irs.gov/pub/irs-soi/{workbook.name}"
+    assert entry["sha256"] == sha256
+    assert entry["size_bytes"] == size == workbook.stat().st_size
+    assert entry["storage"]["r2"]["key"].endswith(f"/2023/{sha256}/{workbook.name}")
+
+
+@pytest.mark.parametrize("package_id", sorted(PACKAGES))
+def test_each_fact_is_its_publisher_cell(built, package_id):
+    _, _, workbook, sha256, _, sheet, expected = PACKAGES[package_id]
+    facts = {fact.source_record_id: fact for fact in built[package_id]}
+
+    assert set(facts) == {f"irs_soi.ty2023.{suffix}" for suffix in expected}
+    for suffix, (address, scale) in expected.items():
+        fact = facts[f"irs_soi.ty2023.{suffix}"]
+        assert fact.value == _cell(workbook, sheet, address) * scale, suffix
+        assert fact.period.type == "tax_year"
+        assert fact.period.value == 2023
+        assert fact.source.source_sha256 == sha256
+        assert fact.source.vintage == "tax_year_2023"
+        assert fact.geography.id == "0100000US"
+        if fact.measure.legal_vintage is not None:
+            assert fact.measure.legal_vintage == "tax_year_2023"
+    for record_id, value in PUBLISHED.items():
+        if record_id.startswith(tuple(f"irs_soi.ty2023.{s}" for s in expected)):
+            assert facts[record_id].value == value
+
+
+def test_state_totals_equal_the_historic_table_2_us_row(built):
+    """Differential across two publisher files: the TY2023 United States
+    workbook and the TY2023 Historic Table 2 CSV (US, AGI stub 0) report the
+    same totals."""
+    csv_path = DATA / "historic_table_2" / "23in55cmcsv.csv"
+    if not csv_path.exists():
+        pytest.skip("23in55cmcsv.csv arrives with chronicle#291")
+    with csv_path.open(newline="", encoding="utf-8-sig") as handle:
+        us = next(
+            row
+            for row in csv.DictReader(handle)
+            if row["STATE"] == "US" and row["AGI_STUB"] == "0"
+        )
+    by_measure = {
+        fact.source_record_id.split(".")[4]: fact.value
+        for fact in built["soi-state-2023"]
+    }
+    variables = {
+        "return_count": ("N1", 1),
+        "adjusted_gross_income": ("A00100", 1000),
+        "eitc_three_or_more_children_returns": ("N59664", 1),
+        "eitc_three_or_more_children_amount": ("A59664", 1000),
+    }
+    for measure, (variable, scale) in variables.items():
+        assert by_measure[measure] == int(us[variable].replace(",", "")) * scale
+
+
+def test_every_published_spot_value_belongs_to_a_package():
+    ids = {
+        f"irs_soi.ty2023.{suffix}"
+        for package in PACKAGES.values()
+        for suffix in package[6]
+    }
+    assert set(PUBLISHED) <= ids
+
+
+@pytest.mark.parametrize("build_year", [2022, 2024])
+def test_building_at_another_year_does_not_relabel_the_2023_files(built, build_year):
+    for package_id in PACKAGES:
+        _, other = _build(package_id, build_year)
+        assert sorted((f.source_record_id, f.value, f.period.value) for f in other) == (
+            sorted(
+                (f.source_record_id, f.value, f.period.value) for f in built[package_id]
+            )
+        )
+
+
+def _declarations(package_id: str) -> dict:
+    path = PACKAGES_ROOT / SOURCE_PACKAGE_ALIASES[package_id] / "source_package.yaml"
+    return yaml.safe_load(path.read_text())
+
+
+@pytest.mark.parametrize("package_id", sorted(PACKAGES))
+def test_packages_mirror_their_2022_twins(package_id):
+    """Differential: apart from year labels, file locations and the cell
+    positions the TY2023 layout moved, the declarations equal TY2022's."""
+    text = (
+        PACKAGES_ROOT / SOURCE_PACKAGE_ALIASES[package_id] / "source_package.yaml"
+    ).read_text()
+    assert "{year}" not in text
+
+    def normalised(spec: dict) -> dict:
+        spec = json.loads(json.dumps(spec))
+        for key in ("package_id", "label"):
+            spec.pop(key)
+        for key in (
+            "vintage",
+            "extracted_at",
+            "artifact_year",
+            "manifest",
+            "resource_directory",
+        ):
+            spec["artifact"].pop(key, None)
+        for record_set in spec["record_sets"]:
+            for key in (
+                "record_set_id",
+                "source_record_id_prefix",
+                "period",
+                "sheet_name",
+            ):
+                record_set.pop(key)
+            for row in record_set["rows"]:
+                row.pop("row_number")
+                for guard in row.get("guard_cells", ()):
+                    guard.pop("row", None)
+                    guard.pop("column", None)
+            for measure in record_set["measures"]:
+                for key in (
+                    "legal_vintage",
+                    "expected_column_header_row",
+                    "expected_column_header",
+                ):
+                    measure.pop(key, None)
+        return spec
+
+    twin = PACKAGES[package_id][0]
+    assert normalised(_declarations(package_id)) == normalised(_declarations(twin))
+
+
+@pytest.mark.parametrize("package_id", sorted(PACKAGES))
+def test_source_suite_passes_agent_acceptance(package_id, tmp_path):
+    suite = build_source_suite(package_id, tmp_path / package_id, year=2023)
+
+    assert suite.agent_acceptance.valid
+    assert suite.agent_acceptance.counts["row_semantic_error_count"] == 0
diff --git a/tests/test_chronicle_source_cells.py b/tests/test_chronicle_source_cells.py
index 1d4603f6..648d5b0b 100644
--- a/tests/test_chronicle_source_cells.py
+++ b/tests/test_chronicle_source_cells.py
@@ -421,3 +421,111 @@ def test_source_cells_from_xlsx_rejects_a_sheet_the_workbook_does_not_carry():
             artifact,
             sheets=("UKPC", "Renamed_by_publisher"),
         )
+
+
+def _test_artifact() -> SourceArtifactMetadata:
+    return SourceArtifactMetadata(
+        source_name="irs_soi",
+        source_table="test",
+        source_file="test.xlsx",
+        url="https://example.test/test.xlsx",
+        vintage="test",
+        sha256="abc123",
+        size_bytes=10,
+        extracted_at="2026-09-27",
+        extraction_method="test",
+    )
+
+
+def _workbook_bytes(data_extent: tuple[int, int], merged: list[str]) -> bytes:
+    workbook = openpyxl.Workbook()
+    sheet = workbook.active
+    for row in range(1, data_extent[0] + 1):
+        for column in range(1, data_extent[1] + 1):
+            sheet.cell(row=row, column=column, value=row * 100 + column)
+    for cell_range in merged:
+        sheet.merge_cells(cell_range)
+    buffer = BytesIO()
+    workbook.save(buffer)
+    return buffer.getvalue()
+
+
+def test_source_cells_from_xlsx_ignores_a_row_merged_to_the_sheet_edge():
+    """23in54us.xlsx (TY2023) merges its footnote row A179:XFD179. openpyxl
+    fills all 16,384 columns with MergedCell placeholders, so the used range
+    grew to ~70 million cells; the parse now stops at the file's own cells."""
+    content = _workbook_bytes((4, 5), ["A6:XFD6"])
+
+    cells = source_cells_from_xlsx(content, _test_artifact())
+
+    assert {(cell.row_number, cell.column_number) for cell in cells} == {
+        (row, column) for row in range(1, 7) for column in range(1, 6)
+    }
+    assert {cell.address: cell.raw_value for cell in cells}["C2"] == 203
+
+
+def test_xlsx_used_range_bounds_equal_openpyxl_without_an_edge_merge():
+    """Invariant: a sheet with no merged range reaching column XFD or the last
+    row keeps openpyxl's (max_row, max_column), whether its merged ranges sit
+    inside the data or run past it; with one, only its placeholders stop
+    counting."""
+    from chronicle.sources.cells import _xlsx_used_range_bounds
+
+    layouts = [
+        ((rows, columns), merged)
+        for rows in (1, 3)
+        for columns in (1, 4)
+        for merged in (
+            [],
+            ["A1:B1"],
+            [f"A{rows + 2}:F{rows + 2}"],
+            [f"B{rows + 1}:B{rows + 4}"],
+            ["C2:H9"],
+        )
+    ]
+    for extent, merged in layouts:
+        sheet = openpyxl.load_workbook(BytesIO(_workbook_bytes(extent, merged))).active
+        assert _xlsx_used_range_bounds(sheet) == (sheet.max_row, sheet.max_column), (
+            extent,
+            merged,
+        )
+
+        edge = openpyxl.load_workbook(
+            BytesIO(_workbook_bytes(extent, [*merged, "A20:XFD20"]))
+        ).active
+        assert edge.max_column == 16_384
+        assert _xlsx_used_range_bounds(edge) == (
+            max(sheet.max_row, 20),
+            sheet.max_column,
+        ), (extent, merged)
+
+
+def test_xlsx_used_range_bounds_ignore_a_column_merged_to_the_last_row():
+    """The row edge works like the column edge. A stub stands in for the
+    sheet: openpyxl would materialise over a million MergedCells for a real
+    A1:A1048576 merge."""
+    from types import SimpleNamespace
+
+    from openpyxl.cell.cell import MergedCell
+    from openpyxl.worksheet.cell_range import CellRange
+
+    from chronicle.sources.cells import _xlsx_used_range_bounds
+
+    real = openpyxl.Workbook().active
+    cells = {
+        (row, column): real.cell(row=row, column=column)
+        for row in (1, 2, 3)
+        for column in (1, 2)
+    }
+    cells.update(
+        {(row, 4): MergedCell(real, row=row, column=4) for row in range(5, 60)}
+    )
+    cells[(4, 4)] = real.cell(row=4, column=4)  # the merge's anchor
+    sheet = SimpleNamespace(
+        merged_cells=SimpleNamespace(ranges=[CellRange("D4:D1048576")]),
+        _cells=cells,
+        max_row=1_048_576,
+        max_column=4,
+    )
+
+    assert _xlsx_used_range_bounds(sheet) == (4, 4)
diff --git a/tests/test_chronicle_source_package.py b/tests/test_chronicle_source_package.py
index 6c7963b5..fd9083dd 100644
--- a/tests/test_chronicle_source_package.py
+++ b/tests/test_chronicle_source_package.py
@@ -1657,6 +1657,20 @@ def test_national_soi_source_package_aliases_validate_fixture_counts():
             "source_record_count": 2,
             "source_region_count": 1,
         },
+        "soi-ira-traditional-contributions-2023": {
+            "record_set_count": 1,
+            "row_count": 1,
+            "measure_count": 2,
+            "source_record_count": 2,
+            "source_region_count": 1,
+        },
+        "soi-ira-roth-contributions-2023": {
+            "record_set_count": 1,
+            "row_count": 1,
+            "measure_count": 2,
+            "source_record_count": 2,
+            "source_region_count": 1,
+        },
     }
 
     for package_id, counts in expected_counts.items():