A Digital Humanities Analysis of Poe's Influence on Doyle
Poe: "The Purloined Letter" (1844)
Doyle: "A Scandal in Bohemia" (1891)
Research Overview
"Dupin was better. Dupin was decidedly smart."
— Doyle's manuscript notes, later revised to "Dupin was a very inferior fellow"
Methodology: This analysis uses computational text analysis to quantify literary influence across six dimensions: dialogue patterns, character proximity networks, narrative structure positioning, character attributes, spatial organization, and sentence-level stylometrics. Click on any metric below for detailed data and methodology.
88%
Narrative Structure Similarity
Click for details →
85.7%
Character Attribute Overlap
Click for details →
56.8%
Stylometric Similarity
Click for details →
Narrative Structure Analysis
We identified 7 key story elements and measured their position (0.0 = beginning, 1.0 = end) in each text. Similarity = 100% minus mean positional difference.
Story Element
Poe Position
Doyle Position
Match
Crime/Problem Introduction
0.455
0.406
✓ Close
Detective Observation
0.479
0.493
✓ Very Close
Resolution Revealed
0.525
0.511
✓ Very Close
Police/Official Bafflement
0.478
0.075
✗ Divergent
Detective Explains Method
0.65
0.72
✓ Close
The 88% similarity indicates both authors structure their detective stories with remarkably similar narrative beats, except for where official failure is revealed.
Character Attribute Overlap
We extracted 50-word context windows around character mentions and identified 8 attribute categories using keyword matching. Overlap = shared attributes / total attributes.
Attribute
Dupin
Holmes
Shared?
Observant
4 instances
20 instances
✓
Intellectual
2 instances
5 instances
✓
Eccentric
4 instances
2 instances
✓
Analytical
3 instances
4 instances
✓
Reclusive
3 instances
2 instances
✓
Physically Distinctive
2 instances
3 instances
✓
Aristocratic
5 instances
1 instance
Partial
Encyclopedic Knowledge
1 instance
8 instances
Partial
Stylometric Analysis
Sentence-level metrics measuring prose style. Similarity calculated as normalized inverse difference across 6 metrics.
Metric
Poe
Doyle
Diff
Mean Sentence Length
34.2 words
23.1 words
+11.1
Max Sentence Length
296 words
104 words
+192
Parentheticals/1000 words
16.5
2.6
+13.9
Semicolons (total)
82
14
+68
Complex Sentence Rate
25.8%
13.0%
+12.8%
Type-Token Ratio
0.245
0.223
+0.022
The low 56.8% similarity reflects Doyle's deliberate modernization of prose style for magazine serialization.
Key Finding: Doyle adopted Poe's deep structure (narrative architecture, character types, relationship topology) while adapting surface features (prose style, spatial organization, narrator agency) for commercial appeal.
Similarity Breakdown
Click any metric to see methodology and detailed data:
Narrative StructureClick for details88%
88%
What it measures: Position of key story elements (crime introduction, detective observation, resolution, explanation) within the narrative arc.
Method: Pattern matching to detect 7 story beats; record position as percentage through text; calculate mean positional difference.
Interpretation: Both authors place crime ~40-45% in, observation ~48%, and resolution ~51-52%. This tight clustering indicates Doyle internalized Poe's narrative formula.
Character AttributesClick for details85.7%
85.7%
What it measures: Overlap in character traits between analogous figures (Dupin/Holmes, Narrator/Watson, etc.).
Method: Extract 50-word windows around character mentions; match against 8 attribute lexicons (observant, intellectual, eccentric, etc.); calculate Jaccard similarity.
Interpretation: Dupin and Holmes share 6 of 7 core attributes. The primary differences are aristocratic emphasis (Poe) vs. encyclopedic knowledge (Doyle).
Relationship TopologyClick for details75%
75%
What it measures: Similarity in character relationship networks based on textual proximity.
Method: Count character co-occurrences within 50-word sliding windows; build weighted network graph; compare edge distributions between analogous character pairs.
Interpretation: Both texts have detective-narrator as strongest bond, narrator as central hub. Doyle's network is more distributed; Poe's clusters tightly around Dupin.
Prose Style (Stylometrics)Click for details56.8%
56.8%
What it measures: Sentence-level prose characteristics including length, complexity, and punctuation patterns.
Surprising Finding: Poe's story is 88.5% dialogue — more than Doyle's 62.2%. But they serve different functions: philosophical dialogue (Poe) vs. action narration (Doyle).
Methodology: Dialogue ratio = characters within quotation marks / total characters. Verbs were classified into cognitive (thinking/perceiving) and action (physical movement) categories using a curated lexicon of 50+ verbs per category.
Text Composition: Dialogue vs. Narration
Poe: 88.5% Dialogue
88.5%
Characters talk about thinking. Dupin lectures on ratiocination; the Prefect reports failures verbally. All key events happen offstage and are reported through speech.
Doyle: 62.2% Dialogue
62.2%
Narrator describes action. Watson follows suspects, witnesses events, describes disguises visually. Scenes unfold before the reader's eyes.
Action verbs indicate physical activity. Higher ratio = more narrated movement and events.
Poe's Top Action Verbs
take8
leave5
enter3
sit3
walk2
seize2
Doyle's Top Action Verbs
come28
go22
take19
run12
walk11
drive8
Interpretation: Doyle's higher action verb ratio (0.63 vs 0.25) confirms the shift from armchair to street-level detection. Watson narrates movement through London; Poe's narrator sits and listens.
Character Proximity Networks
What the numbers mean: Edge numbers show co-occurrence count—how many times two characters appear within 50 words of each other. Higher count = stronger textual relationship. Click any character node for detailed information.
Methodology: Character positions identified via regex pattern matching (including pronouns for narrators). Co-occurrence counted using 50-word sliding windows. Network edges weighted by count; node size by total mentions.
Poe: The Purloined Letter
Dupin
Narrator
Minister D
Royal Lady
Prefect
Doyle: A Scandal in Bohemia
Holmes
Watson
Irene Adler
King
Norton
Character Role Alignment
Role
Poe Character
Mentions
Doyle Character
Mentions
Detective
Dupin
34
Holmes
58
Narrator
Narrator
199
Watson
396
Antagonist
Minister D
20
Irene Adler
28
Client
Royal Lady
13
King
44
Supporting
Prefect
30
Godfrey Norton
12
Detective
Narrator
Antagonist
Client
Supporting
Narrative Structure: 88% Similarity
Story Beat Alignment: Both authors position crime introduction, detective observation, and resolution at nearly identical points in the narrative arc—evidence that Doyle internalized Poe's formula.
Methodology: Pattern matching identified 7 story elements using keyword triggers. Position recorded as decimal (0.0 = start, 1.0 = end). Similarity = 100% - (mean positional difference × 100).
Story Element Positioning
Start (0.0)Middle (0.5)End (1.0)
●●Crime Intro
●●Detective Observes
●●Resolution
●●Police Baffled ⚠️
●●Explanation
● Poe● Doyle
Story Element
Poe Position
Doyle Position
Difference
Crime Introduction
0.455
0.406
0.049 ✓
Detective Observes
0.479
0.493
0.014 ✓
Resolution
0.525
0.511
0.014 ✓
Police Baffled
0.478
0.075
0.403 ✗
Detective Explains
0.65
0.72
0.07 ✓
Notable Divergence: "Police Baffled" appears at 0.478 in Poe (midpoint) vs 0.075 in Doyle (opening). Doyle front-loads the problem setup to create space for active investigation scenes.
Spatial & Narrative Space Analysis
Fundamental Difference: Poe's nested narrative structure contains all spaces within Dupin's library—events are reported, not shown. Doyle's narrative moves physically through London—Watson witnesses action in real time.
Methodology: Named Entity Recognition (spaCy) extracted place names. Each location classified as: Real (verifiable), Conjectured (fictional but mapped to likely inspiration based on scholarly research), or Reported (mentioned but never visited in narrative). Movement scenes counted by action verbs + location co-occurrence.
Narrative Space Structure
Poe: Nested/Contained Space
All action contained within one room; other spaces exist only in speech
📚 Dupin's Library (No. 33 Rue Dunôt)
Primary setting — entire story takes place here
Minister's Hotel REPORTED
Where letter hidden — described by Prefect, never visited
Royal Boudoir REPORTED
Where theft occurred — mentioned, never shown
Vienna REPORTED
Where Minister wronged Dupin — backstory only
0 movement scenes — armchair detection
Doyle: Linear/Mobile Space
Watson physically moves through London; reader witnesses action
🏠 221B Baker Street
Opening: Watson visits Holmes
↓ cab ride
🏘️ Briony Lodge, St. John's Wood
Surveillance, disguise, smoke-rocket scheme
↓ follows cab
⛪ Church of St. Monica
Watson witnesses wedding
↓ walks
🏨 Langham Hotel
King of Bohemia's residence
8 movement scenes — street-level detection
Interactive Location Map
📍 Poe's Paris (1844)
⬥ Rue Dunôt, Faubourg St-Germain
Dupin's library — entire story takes place here
CONJECTURED → Rue du Bac area
● Faubourg Saint-Germain
Aristocratic Left Bank district — Dupin's neighborhood
REAL — still exists
⬥ Minister's Hotel
Where the letter is hidden — only described, never shown
REPORTED ONLY
⬥ Rue Morgue
Reference to earlier Dupin story — symbolic name
FICTIONAL — "Street of the Dead"
● Vienna
Where Minister D— wronged Dupin
REAL — backstory reference
Note: Poe never visited Paris. His geography is atmospherically evocative but invented.
📍 Doyle's London & Europe (1891)
● 221B Baker Street
Holmes's residence — opening and closing scenes
REAL — street exists (221B fictional in 1891)
⬥ Briony Lodge, Serpentine Avenue
Irene Adler's residence — surveillance scenes
CONJECTURED → Langford Place, St. John's Wood
⬥ Church of St. Monica
Wedding scene — Watson witnesses
CONJECTURED → Edgware Road area
● Langham Hotel
Where the King stays — still exists today
REAL — Doyle dined here with Oscar Wilde (1889)
● Regent Street
Gross & Hankey's jewelers — Holmes in disguise
REAL — major shopping street
● Edgware Road
Route to church — mentioned in dialogue
REAL — major London road
● St. John's Wood
Affluent area where Irene Adler lives
REAL — known for private villas
● Inner Temple
Godfrey Norton's legal chambers
REAL — Inns of Court
● Bohemia (Prague)
Kingdom of the client — now Czech Republic
REAL — historical region
⬥ Cassel-Felstein
Part of King's title — fictional German duchy
CONJECTURED → Kassel (spelled "Cassel" until 1926)
● Warsaw
Where King met Irene Adler years ago
REAL — backstory reference
● Holland
Reference to Holmes's case for Dutch royal family
REAL — mentioned case
● Odessa
Reference to Trepoff murder case
REAL — Ukraine
Spatial Metrics Comparison
Metric
Poe
Doyle
Settings Shown (reader visits)
1
5+
Settings Reported Only
3
0
Named Locations (total)
5
14
Movement/Travel Scenes
0
8
Verifiable Real Locations
2
9
Conjectured Locations
3
3
Geographic Scope
Paris + Vienna
London + 5 European locations
Stylometric Divergence: 56.8% Similarity
Greatest Divergence: Prose style shows the lowest similarity of any metric. Doyle deliberately modernized Poe's ornate style for magazine serialization and mass readership.
Methodology: Sentence boundaries detected using spaCy. Metrics calculated: mean/max sentence length, parenthetical count (text in parentheses or dashes), semicolon/em-dash frequency, complex sentence rate (sentences with 2+ clauses), type-token ratio (unique words / total words).
Metric
Poe
Doyle
Difference
Mean Sentence Length
34.2 words
23.1 words
+11.1
Max Sentence Length
296 words
104 words
+192
Parentheticals/1000 words
16.5
2.6
+13.9
Semicolons (total)
82
14
+68
Em-dashes (total)
67
12
+55
Complex Sentence Rate
25.8%
13.0%
+12.8%
Type-Token Ratio
0.245
0.223
+0.022
Poe's Prose Style
"The measures adopted were not only the best of their kind, but carried out to absolute perfection. Had the letter been deposited within the range of their search, these fellows would, beyond a question, have found it."
Long, complex sentences (avg 34.2 words)
Heavy use of parenthetical insertions
Abundant semicolons and em-dashes
Latinate, elevated vocabulary
Multiple subordinate clauses
Doyle's Prose Style
"The door opened. A lady came out. She was a very handsome woman."
Short, punchy sentences (avg 23.1 words)
Conversational, direct tone
Simple clause structures
Accessible, everyday vocabulary
Visual, cinematic descriptions
Commercial Strategy: Writing for The Strand Magazine, Doyle needed prose suited for commuters reading in short bursts. He modernized the detective story not by changing its structure but by streamlining its prose.