Same source. Different facts or readings.
Follow each saved answer to its evidence.
Loading saved comparisons…
Explore the text, its annotations and the evidence behind the reader.
Loading the corpus atlas…
Surah 24 · 64 verses
النور
Select an underlined passage to open its annotation. The source text is Arabic; the interface is English.
وَٱلَّذِينَ يَرۡمُونَ ٱلۡمُحۡصَنَٰتِ ثُمَّ لَمۡ يَأۡتُواْ بِأَرۡبَعَةِ شُهَدَآءَ فَٱجۡلِدُوهُمۡ ثَمَٰنِينَ جَلۡدَةٗ وَلَا تَقۡبَلُواْ لَهُمۡ شَهَٰدَةً أَبَدٗاۚ وَأُوْلَٰٓئِكَ هُمُ ٱلۡفَٰسِقُونَ
The pinned Arabic edition of An-Nur. Enable JavaScript to explore the clause annotations.
Explore how the model of An-Nur responds to different facts and named readings.
Text, recorded classification and an executed answer are three different kinds of evidence.
Books come from historical Git trees identified by qualified farm transfer records. Each book retains its accepted HEAD, source and model fingerprints, dependency snapshot and recorded evidence. Ordinary builds use a frozen release; ongoing farm changes enter only through an explicit release refresh.
Text, normative qualification, named readings and attributed grades remain separate. Scenario evidence is recorded, not rerun by the reader. No English scripture translation is included.
The reading catalogue starts from the repository's pinned Tanzil Uthmani text, version 1.1. Annotations use the exact Arabic fragment of their own corpus edition. An-Nur uses a pinned Wikisource copy of the Medina mushaf (Hafs); its digital spelling differs from Tanzil.
No English translation of the Quran text is included in this release. English interface explanations and model identifiers are not a translation of the scripture.
Tanzil Project · Text licence and notice · Pinned Tanzil XML
Clause classifications come from the corpus's QDC-1 review inventories: direct norm, norm under a reading, non-normative content, external owner or unresolved work. A linked rule alone does not establish a normative position.
Recorded reviews and scenario evidence are displayed with their source. They are not presented as a new acceptance of the corpus. Missing or changed source fingerprints are shown explicitly.
Historical address and scope remain those recorded in the model. The reader does not infer modern applicability.
Each setting selects a registered case. Its answers were calculated by the Rust engine, checked against the collection's authored expectations and replayed. The independent Python evaluator was not run.
“Not established” means the model does not establish an answer on the facts provided. It does not mean “false”. A required reading remains visible until a named alternative is selected.
This is a research view of a model, not a fatwa or tafsir.