Within-Surah Verse-Length Outlier at Quran 37:102

Abstract

QMR-012 examines the length of Quran 37:102 relative to the other verses of its own surah.

Under the fixed DATA-001 counting conventions, Quran 37:102 contains:

• 100 letters
• 28 words

For every verse in the corpus, its length was standardized relative to the distribution of verse lengths inside its own surah.

Using the population standard deviation, Quran 37:102 has:

• letter-length Z-score: 8.987301774193 in BODY;
• word-count Z-score: 9.918243692516 in BODY.

For both measurements, Quran 37:102 is the unique rank 1 / 6,236 observation.

The same rank is obtained under RAW, with only negligible changes in the standardized values.

This result is a descriptive ranking of the fixed corpus. QMR-012 assigns no inferential p-value to the observation.

Deterministic Result

DATA-001 contains:

• 114 surahs
• 6,236 verses

For each verse, two measurements were computed:

1. Arabic-letter count;
2. whitespace-delimited word count.

Each measurement was then standardized within the verse’s own surah:

Z = (verse length - surah mean) / surah standard deviation

The resulting BODY values for Quran 37:102 are:

In both cases:

• no verse has a higher within-surah population Z-score;
• there is exactly one observation at the maximum value;
• the maximum is Quran 37:102.

Therefore the result is not merely a statement that 37:102 is long in absolute terms. It is specifically the highest standardized positive verse-length departure from its own surah among all verses in DATA-001 under these two counting measures.

Surah 37 Context

Surah 37 contains 182 verses.

Letters

Quran 37:102 contains 100 letters.

The second-highest distinct letter count in the surah is:

48

The gap is therefore:

100 - 48 = 52 letters

The maximum-to-mean ratio is:

4.667863554758

Quran 37:102 is the sole 100-letter verse in the surah.

Words

Quran 37:102 contains 28 words.

The second-highest distinct word count in the surah is:

12

The gap is:

28 - 12 = 16 words

The maximum-to-mean ratio is:

5.891329479769

Quran 37:102 is also the sole 28-word verse in the surah.

Ranking Among Surah Maxima

A second comparison uses one maximum observation from each of the 114 surahs.

For Surah 37:

This distinction matters.

Quran 37:102 is first under standardized within-surah departure, but it is not first under every possible measure of extremeness.

The result is therefore stated specifically in terms of the standardized statistic rather than generalized into a broader claim.

Sensitivity Analysis

Population vs sample standard deviation

The ranking is unchanged when the sample standard deviation is used.

For BODY:

Quran 37:102 remains rank 1 under both versions.

Leave-one-out standardization

A further control removes the tested verse before estimating the mean and standard deviation of its surah.

For BODY:

• letters, leave-one-out Z: 12.077208671905
• words, leave-one-out Z: 14.679208925274

The rankings become:

• letters: positive rank 4 / 6,236
• words: positive rank 3 / 6,234

The leave-one-out result therefore does not preserve rank 1.

This is an important limitation rather than a contradiction. Several higher leave-one-out values occur in much smaller surahs, where removing one observation can greatly change the estimated dispersion.

Quran 37 contains 182 verses, so its result is not driven by an extremely small within-surah sample.

RAW / BODY Robustness

Quran 37:102 itself is unaffected by the BODY transformation.

Its count remains:

• RAW: 100 letters, 28 words
• BODY: 100 letters, 28 words

The standardized values are also very close.

Letters

• RAW population Z: 8.987339829562
• BODY population Z: 8.987301774193

Words

• RAW population Z: 9.939281285416
• BODY population Z: 9.918243692516

The global rank remains uniquely first for both letters and words in both modes.

Statistical Interpretation

QMR-012 is a descriptive corpus-ranking result.

No generative probability model is asserted for the distribution of within-surah maxima.

Accordingly, QMR-012 reports:

no inferential p-value

and does not convert the large numerical Z values into normal-tail probabilities.

The within-surah verse-length distributions are not assumed to be normally distributed, and the fact that a standardized value is numerically large does not by itself establish an inferential probability statement.

The Z-score is used here as a common scale for comparing departures from different surah-specific means and dispersions.

Prior Art

The broader observation that Quran 37:102 is unusually long is previously known.

Nicolai Sinai discusses Q 37:102 as exhibiting a pronounced increase in verse length relative to its literary environment. In his own transcription-based counting system, he reports a mean verse length of 31.2 transcription letters for Surah 37 and a length of 143 transcription letters for verse 102.

Those figures are not directly interchangeable with DATA-001 because the counting conventions differ.

Behnam Sadeghi’s 2011 stylometric research also uses quantitative verse length as a stylistic variable and discusses mean verse length and dispersion across Quranic textual groupings.

These precedents establish prior art for:

• quantitative Quranic verse-length analysis;
• comparison of verse lengths with surrounding textual distributions;
• recognition of Quran 37:102 as an unusually long verse.

In the sources examined for QMR-012, we did not locate the exact exhaustive DATA-001 claim tested here: ranking all 6,236 verses by their within-surah standardized length independently for both letter count and word count.

Because the underlying length anomaly at 37:102 was already identified in earlier scholarship, QMR-012 is classified conservatively as:

previously known

and:

discovery: false

The contribution of QMR-012 is reproducible corpus-wide quantification under a fixed modern corpus and explicit counting rules.

Historical Reconstruction

REC-012 began from a recovered 2025 record describing Quran 37:102 using several numerical claims, including:

• 102 letters;
• 30 words;
• a score of 8,372;
• a Z-score of approximately 1.512;
• a rank among 104 so-called auto-references.

Those values do not reproduce together under one fixed DATA-001 convention.

Under DATA-001:

• letter count = 100
• whitespace-delimited word count = 28
• strict Abjad score = 8,372

The historical Z-score was found to track DATA-001 strict-Abjad values very closely across the recovered top-five table, but the complete historical scoring framework and the claimed population of 104 cases were not reconstructed.

A separate Tanzil Uthmani forensic corpus showed that 37:102 can reach a count of 102 when Unicode Arabic letters are supplemented by five U+0670 superscript-alef marks.

That forensic reconstruction is not used for QMR-012.

QMR-012 therefore separates the recovered historical record from the modern result and retains only the latter as the public observation.

Limitations

The main limitations are:

1. The result depends on the explicit DATA-001 text representation and counting conventions.
2. Standard Z-scores provide a standardized descriptive scale; they do not establish a probability model.
3. Rank 1 does not survive the leave-one-out transformation, although 37:102 remains among the most extreme observations.
4. Different Quranic orthographies and tokenization systems can produce different raw letter and word counts.
5. The analysis was motivated by a previously identified historical target, so the target was not selected independently before inspection of the corpus.
6. Prior-art searching cannot prove the absolute absence of an earlier publication making the exact same exhaustive ranking claim.

For these reasons, QMR-012 is published as an E1 observation, not as an inferential statistical result.

Reproducibility

Canonical corpus:

DATA-001

Corpus SHA-256:

228df2a717671aeb9d2ff573002bd28d6b3f973f4bc7153554e3a81663d67610

Analysis:

analysis/qmr-012.mjs

Deterministic result artifact:

artifacts/qmr-012/qmr-012-data001-report.json

Artifact SHA-256:

B0C3AD264DA2D2B48C277BABC817C30979D7122C38FA090891929C38159854FE

Artifact schema:

corancodex.qmr-012.v1

The artifact is required to equal a fresh execution of the analysis script against the canonical DATA-001 corpus.

References

• Nicolai Sinai, The Qur’an: A Historical-Critical Introduction, Edinburgh University Press, 2017.
• Behnam Sadeghi, “The Chronology of the Qur’an: A Stylometric Research Program”, Arabica 58 (2011), pp. 210-299, DOI: 10.1163/157005810X529692.
• Tanzil Project, Quran Text — Simple Clean — Version 1.1.