Can the selected Scope 2 records support a method-specific comparison
No method-specific comparison can be supported by the retained classifications alone. All 51 selected Scope 2 observations in this frozen audit remain method-unspecified: none is explicitly classified as location-based or market-based. That finding concerns a defined historical dataset, not what every issuer disclosed elsewhere or which method it actually used.
This is a historical research edition based on Nitium’s 0.12.15-local1293-balamines snapshot, frozen on 6 October 2026. It preserves 51 observations from 22 issuers and 30 source editions. Reporting periods end in March 2023 through March 2026. The selected pool is purposive, not a representative survey or a measure of reporting compliance.
Of the 51 observations, 39 appear in the separately verified historical 1,040-record public projection; 12 were local-only at the freeze. The full audit cannot be reproduced from that public download alone. Its frozen inputs, stable identifiers and review references are retained for checking, with no new company-data credit.
What the audit actually tests
The inclusion rule selects accepted observations in the greenhouse-gas collection whose metric identifier begins with GHG_SCOPE2. It excludes Scope 1, Scope 3, combined-scope ratios, held cells and superseded source-history alternatives. Each selected observation is counted once. Duplicate identifiers would stop the audit rather than inflate the denominator.
The classification comes from accepted row metadata, not a new judgement about every page of every report. The audit retains each record’s period, boundary, original unit, method note, source location and separate review reference. One report may supply a current value and a comparative; one issuer may contribute several years. Observations, reports and companies are therefore different counting units.
The result is 0 explicit location-based records, 0 explicit market-based records and 51 method-unspecified records. These are counts of classifications. Zero does not mean zero emissions, absence of a method in company practice or proof that a method statement cannot be found in other evidence. No emissions quantities are summed, converted or ranked.
Four source-note tests that keep questions separate
Mazagon Dock’s 2024–2025 Scope 2 note names a CEA edition. The retained accounting-method state remains unspecified; a factor reference does not itself supply an explicit method label.
eClerx’s 2024–2025 table labels Scope 2 in tonnes of CO2 and Scope 1 in tonnes of CO2 equivalent. Preserve that unit distinction separately from the unresolved accounting-method state.
Dwarikesh’s 2025–2026 Scope 2 note covers corporate offices in Delhi, Maharashtra and Uttar Pradesh. That office boundary cannot be extended to manufacturing plants or turned into an accounting-method designation.
Mrs. Bectors’ 2025–2026 notes identify a grid-factor version, manufacturing coverage and a restated comparative. Those details do not close the retained gap in explicit accounting-method classification.
The source catalogue below gives the exact report links. The four illustrated passages were reread during this preparation, including their relevant notes. They show why “method complete” is too broad a single field: factor provenance, gas-unit identity, activity coverage and accounting-method wording answer different questions. Resolving one does not automatically resolve the others.
Original test: a default that changes the answer
Consider three entirely fictional source records. Record A names a factor reference but has no explicit accounting-method label. Record B expressly labels its quantity location-based. Record C expressly labels its quantity market-based. For this exercise, the task is to preserve what the invented records say; it does not validate whether any underlying calculation complies with a method.
The faithful classification is one unspecified, one explicitly location-based and one explicitly market-based record. Now imagine an export setting that replaces every unspecified value with location-based. Its summary becomes zero unspecified, two location-based and one market-based. The row count still equals three, so a simple completeness check does not detect the change.
No new evidence arrived. The default changed the apparent comparison eligibility by changing the label on Record A. A tidy chart could now suggest that two source records declared the same method when only one did. This is a metadata transformation error, even though the export preserved every record identifier and did not change any reported amount.
An adversarial check compares each output classification with its retained source state and looks specifically for an unspecified-to-explicit transition without new evidence. Keep an analyst’s inference in a separate field with its basis and review status. Never let that field overwrite the attributed source wording or silently change the frozen historical classification.
If later evidence supplies a method statement for Record A, create a dated revision that identifies the exact passage, applicable period and affected record. Keep the earlier unspecified state in history. A justified update and a software default can produce the same visible label, but only the first has a new evidential basis.
A five-step comparison gate
1. Define the output. A descriptive inventory of source-reported quantities and a method-specific benchmark are different products. State the purpose before filtering records. Do not lower the evidence requirement simply because a chart would otherwise contain fewer entries.
2. Preserve the method evidence. Keep the issuer statement, exact source location and period together. Use an explicit unspecified state where the retained evidence does not establish a designation. An absent field should prompt a question, not a convenient default.
3. Check unit and coverage independently. Record the gas label, entities, sites and purchased-energy activities actually covered. A common company name does not establish common operating boundaries. A method label would not, by itself, make different units or populations interchangeable.
4. Keep factor provenance separate. Preserve the provider, edition, unit, geography and row or year markers. A note attached to one cell should not spread across every year in a table. These checks can clarify an input without proving the accounting-method designation.
5. Release a qualified result. Use a record only for the statement its retained evidence supports. Leave it outside a method-specific comparison when the required classification is unresolved. Record the clarification needed and version any later decision; do not portray this holding state as a finding that the issuer’s calculation is wrong.
All issuers and source editions in the frozen count
This catalogue preserves the full historical coverage. Counts are selected observations per issuer. Each report-year link identifies the selected GHG-table PDF page; continuation notes remain part of the record. Current access limits are stated below. Links identify the fixed editions used, not a latest-report search or a company ranking.
Andhra Paper Limited: 2 selected records. 2025 report, PDF 32.
Aurobindo Pharma Limited: 2 selected records. 2025 report, PDF 130; right panel (one-based PDF page).
Avadh Sugar & Energy Limited: 2 selected records. 2026 report, PDF 131.
Balaji Amines Limited: 4 selected records. 2024 report, PDF 53 left; 2025 report, PDF 57 left; 2026 report, PDF 125.
Mrs. Bectors Food Specialities Limited: 4 selected records. 2024 report, PDF 25; 2025 report, PDF 27; 2026 report, PDF 28.
City Union Bank Limited: 2 selected records. 2025 report, PDF 49.
Cochin Shipyard Limited: 3 selected records. 2024 report, PDF 24; 2025 report, PDF 26.
Dwarikesh Sugar Industries Limited: 2 selected records. 2026 report, PDF 245.
eClerx Services Limited: 2 selected records. 2025 report, PDF 25.
Gateway Distriparks Limited: 2 selected records. 2025 report, PDF 19.
JK Lakshmi Cement Limited: 2 selected records. 2025 report, PDF 167 (one-based PDF page).
K.P.R. Mill Limited: 2 selected records. 2024 report, PDF 77; 2025 report, PDF 77.
KRBL Limited: 2 selected records. 2025 report, PDF 34.
Mazagon Dock Shipbuilders Limited: 2 selected records. 2025 report, PDF 39.
NBCC (India) Limited: 2 selected records. 2025 report, PDF 30.
Rashtriya Chemicals and Fertilizers Limited: 2 selected records. 2025 report, PDF 31.
Sandhar Technologies Limited: 2 selected records. 2025 report, PDF 33.
Suprajit Engineering Limited: 2 selected records. 2026 report, PDF 86.
Tega Industries Limited: 4 selected records. 2024 report, PDF 75; 2025 report, PDF 80; 2026 report, PDF 73.
Vardhman Textiles Limited: 2 selected records. 2025 report, PDF 32.
V-Mart Retail Limited: 2 selected records. 2025 report, PDF 30.
West Coast Paper Mills Limited: 2 selected records. 2025 report, PDF 53.
Access and reproduction limits
Preparation check, 15 October 2026: 26 of the exact report PDFs were retrieved. The three Tega URLs returned HTTP 403; West Coast’s current access was not verified. Tega’s original PDFs and page images were not recovered locally. Its dated source hashes, download records, cell fixtures and separate review records remain in the historical evidence package. No fresh Tega page inspection is claimed.
The count is reproducible from the hash-verified frozen observation file by applying the inclusion rule, checking unique identifiers and grouping the retained classifications. It is not a fresh all-pages review of 30 reports. Historical source and review records support provenance but do not establish that a currently inaccessible source remains unchanged. The access limitation stays attached to this edition.
Other reports, newer editions or issuer evidence may resolve individual questions. This study does not search all possible disclosures for all 22 issuers. A later research edition should identify new evidence and changes explicitly instead of silently replacing the historical population. Source checking concerns transcription and context; it does not independently verify underlying emissions or establish issuer endorsement.
The reported greenhouse-gas data hub carries a later public edition. Its coverage does not replace this study’s frozen population. Use each edition’s own selection, source history and qualifications; do not treat different dataset counts as a change in company emissions.
Historical input freeze: 6 October 2026. Preparation: 15 October 2026. Source assets and third-party assurance material are excluded. The five-step gate, fictional test and artwork are original Nitium material. Retained classifications do not establish measurement assurance.
Comments