Selecting Comparison Examples

Selecting comparison examples is not a preliminary housekeeping task. It is one of the decisions that most strongly shapes a comparative conclusion. The examples chosen determine which similarities appear important, which differences become visible and whether the final judgement is supported by evidence or merely looks persuasive.

A useful comparison example is not simply an object, image or record that resembles the collectible under study. Its own identity must be understood, its relevance to the research question must be clear and the differences between it and the target object must be controlled well enough to interpret. A weak comparator can make an ordinary object appear rare, a later reproduction appear early, a restored surface appear original or a mixed set appear factory-complete.

Good selection therefore begins before the side-by-side examination. The collector states the question, defines the comparison population, identifies the variables that could explain difference, chooses complementary examples and records why each example was included. The aim is not to assemble the most convincing-looking group. It is to build a comparison set that another knowledgeable person could understand, repeat and challenge.

Collector scenario

Three close-looking examples do not necessarily make a secure match

A collector buys a boxed item described as an early UK issue. Three online listings show the same front artwork, logo position and general colour. Taken together, they appear to confirm the seller's description.

The comparison weakens when the examples are examined properly. One is a North American release, one is the same physical object resold through a second platform and none shows the rear label, insert or production code. The apparent group of three independent matches is really one regional mismatch, one repeated object and one incomplete image record.

The error was not poor eyesight. It was poor comparator selection. The examples looked relevant before the question, variables and source relationships were controlled.

Begin with the research question

The correct comparator set changes with the task. An example that helps identify a maker may be unsuitable for condition judgement, and a market comparable may reveal very little about production chronology.

What is this object?

Identification

Prefer securely catalogued examples, manufacturer or publisher records, institutional holdings and objects with unambiguous markings. A broad visual match may identify the product family while remaining insufficient for a specific edition, factory or production state.

Which printing, state or regional release is this?

Edition or variant

Use multiple confirmed examples immediately before, within and after the suspected state. Transitional and negative examples are often more informative than a single textbook copy because they show whether a feature is genuinely diagnostic.

Is it contemporary and substantially original?

Authenticity

Compare several documented originals, several known false examples and uncertain objects kept separate from both groups. Resemblance to one presumed original, or failure to match one known fake, is not a complete authentication method.

Do the correct parts belong together?

Completeness

Use original inventories, instructions, period catalogues, opened complete examples and component-level comparisons. A sealed example can support external packaging evidence but may not establish every internal component or prove that a seal is original.

What is manufacture, ageing, damage or intervention?

Condition

Choose examples with comparable materials and, where possible, documented storage or treatment histories. Include untreated and known restored objects so that fading, shrinkage, cleaning, overpainting or replacement are not mistaken for production differences.

Which historical sales are genuinely comparable?

Valuation

Match edition, format, condition, completeness, provenance, sale date, venue and price basis. Asking prices, hammer prices and prices including buyer's premium are different records and should not be merged into one apparent market figure.

A hierarchy of comparison relevance

Comparisons become less direct as they move away from identity. Lower levels can still be valuable, but they must not be allowed to carry the weight of a closer match.

Level 1

Identity matches

The closest available examples: same product, edition, production state, region, factory, material specification and intended contents.

same printingsame mould and factorysame regional issue

Limit: They are normally the strongest direct comparisons, but their own attribution, condition and component history still require scrutiny.

Level 2

Near-identity matches

Examples differing in one or two known respects, such as an adjacent printing, revised insert, neighbouring serial range or changed regional packaging.

old logo versus revised logosame edition with a new insertsame mould with a changed paint scheme

Limit: These are especially useful for isolating one change, provided the remaining differences are stated rather than ignored.

Level 3

Structural analogues

Different objects made through a relevant shared production system, workshop, printer, moulding process or publisher practice.

another title from the same printeranother range using the same factory markanother product made from the same tool

Limit: They can explain manufacturing conventions but cannot automatically establish the exact form of the target object.

Level 4

Contextual comparisons

Objects linked by period, genre, market or collecting practice rather than close production identity.

competing productslater reissuessimilar objects from another maker

Limit: Useful for historical context and broad interpretation, but weak as direct proof of a specific state or feature.

Level 5

Illustrative examples

Objects selected to explain a concept, such as a dramatic fake, a heavily faded surface or a conspicuous restoration.

known counterfeit typographyextreme light damageobvious replacement component

Limit: Illustration teaches recognition; it should not be presented as representative population evidence.

Match the variables that matter

The most informative comparison changes the feature under investigation while keeping as many other variables stable as the evidence allows. Perfect control is rare, so remaining confounders should be named rather than hidden.

Identity and issue

maker, publisher, artist or manufacturer
title, model, catalogue number or production code
edition, printing, mould, serial range, language and region

Chronology

manufacture, publication and release dates
packaging revision, sale, restoration and documentation dates
independent chronological anchors rather than an unchanged copyright date

Manufacturing source

factory, subcontractor, press, tool or mould
locally sourced materials and concurrent suppliers
batch relationships that may imitate chronology or rarity

Materials and construction

substrate, paper, board, plastic, alloy, glaze, ink, paint and adhesive
binding, stitching, fasteners, mould seams and assembly sequence
material differences that may create different ageing patterns

Dimensions and scale

height, width, depth, thickness and weight
component dimensions, margins and spacing
actual measurements rather than estimates from perspective photographs

Condition, completeness and provenance

fading, shrinkage, abrasion, cleaning, trimming, repair and replacement
which parts survive, which are missing and which may be married
whether components remained together and whether intervention is documented

Single-variable principle

Meaning

A useful lead, candidate or image record when its limitations are preserved.

Collector risk

The identity, region, condition, ownership history and originality may all be assumed. A confident seller label is not independent confirmation.

Secure, provisional and circular attribution

Secure

Supported by several independent indicators such as manufacturer documentation, accession records, purchase history, known discovery context, physical evidence and specialist consensus.

Provisional

Likely correct but dependent on incomplete evidence. It can still be useful when its uncertainty is preserved and it is not allowed to define the group by itself.

Circular

Identified from a claim that is repeated through sellers, guides, databases and collectors without an independent evidential foundation. Repetition increases visibility, not reliability.

Build a comparison set, not a shrine to one exemplar

One object rarely establishes the normal form of a mass-produced collectible. A stronger set assigns different examples different jobs instead of asking one ideal specimen to answer every question.

Representative examples

Show the form that appears normal within a group: usual dimensions, typical contents, common manufacturing variation and expected wear.

Diagnostic examples

Carry repeatable features that distinguish one group from another, such as a revised printer's mark, mould modification, corrected text or changed safety label.

Boundary examples

Sit near a transition: old packaging with new contents, over-stickered codes, mixed production features or overlapping serial ranges. They reveal how change actually occurred.

Negative controls

Are securely known not to belong to the proposed group. They test whether a supposed diagnostic feature also occurs where it should not.

Anomalous examples

Do not fit the current pattern. They should be retained and investigated rather than deleted because they complicate the preferred explanation.

Same-object records

Show the exact object at an earlier date. They are not independent examples of the type, but may be the best evidence for restoration, component change, provenance revision or price history.

Positive and negative controls

A positive control is known to possess the feature being tested. A negative control is known not to possess it. Both are needed because a feature may be common across several states even when it appears in every suspected early example.

Ask both: is the feature present where expected, and is it absent where it should not occur?

How many examples are enough?

Three examples may begin to show repetition in a narrow feature. Five to ten independent examples can reveal variation, subgroups and outliers. Common mass-produced objects may require dozens or hundreds of observations to estimate frequency or chronology.

There is no universal minimum. Rare objects may permit only a small sample, but a small sample should produce a correspondingly limited claim.

Control condition, intervention and image limitations

The object in front of the camera is not necessarily the object that left the factory. Comparison must separate manufacture from ageing, damage, treatment and later assembly.

Condition can imitate production difference

faded red can appear orange
yellowed coatings can warm colour
shrinkage can alter dimensions
abrasion can remove fine text
oxidation can change an apparent finish
trimming can imitate a smaller format

Restoration can imitate originality

recolouring can create an unnaturally uniform surface
replacement packaging can make a mixed object look complete
cleaning can remove ageing evidence
overpainting can conceal wear or repair
reassembly can make unrelated parts appear historically associated

Images can create or conceal difference

lighting, white balance and colour profile
perspective, lens distortion and cropping
compression, sharpening and scanner settings
glare, reflection, scale and screen calibration
missing backs, edges, interiors or identifiers

Sampling can distort the apparent population

high-value, unused and visually striking objects survive and circulate more often
ordinary, damaged, inexpensive and regional examples are under-recorded
museum and elite auction samples can exaggerate completeness and condition
many examples may still derive from one batch, owner or copied claim

Research quality and market desirability are not the same

Collectors naturally gravitate toward the cleanest, most complete and most expensive example. Yet a pristine object may have been cleaned, restored, recoloured, reassembled or married with better components. It can be a poor comparator for original surfaces and ageing.

A worn but untouched object may preserve exposed substrate, adhesive residue, ownership marks, construction details and long-standing component associations. The most desirable object is not automatically the most informative research object.

Compare components before declaring a set complete

Composite collectibles should be decomposed before being judged as a whole. Each box, insert, manual, accessory, certificate or wrapper may have its own production and ownership history.

Component identity

production codes and region
materials, printing and dimensions
expected fit and packaging relationship

Historical association

matching wear and compatible ageing
shared ownership marks
provenance showing the parts remained together

Alternative explanations

factory transition
regional mixing
later replacement or owner-created completeness

A defensible selection workflow

A defensible method defines selection criteria before the researcher knows which examples best support the preferred conclusion.

State the question

Write one precise sentence. Replace 'is this the same?' with a testable question about edition, component relationship, originality, condition or market comparability.

Define the comparison population

Specify product, region, date range, edition, component and acceptable evidence sources. This prevents the search drifting toward attractive but irrelevant matches.

List the explanatory variables

Identify chronology, factory, region, material, condition, restoration, completeness and owner intervention before deciding what a difference means.

Set inclusion and exclusion criteria

Decide what an example must show and what makes it unsuitable. Keep excluded examples in a separate log so later researchers can see that they were considered.

Gather broad candidates

Search more widely than the final set. Separate secure, probable, uncertain, disputed and known false examples rather than allowing all candidates to carry equal weight.

Choose complementary roles

Build a set containing representative, diagnostic, boundary, negative and anomalous examples where available. Do not collect only examples that support the preferred answer.

Record consistently

Use the same measurements, image requirements, field names and absence codes across every comparator. 'Not photographed' and 'not visible' are not the same as 'absent'.

Compare before narrating

Populate the matrix first. Then ask which patterns repeat, which conflicts remain and whether one weak example carries too much of the conclusion.

Test alternatives

Ask what else could explain the pattern: ageing, restoration, regional supply, shared batch, mixed components, data error or a counterfeit construction.

State limits and confidence

Record sample size, missing regions, inaccessible objects, uncertain dates, image-only examinations and possible source dependence. Limited evidence should produce limited claims.

Document every comparator

Comparison notes should preserve the object-level evidence even if terminology or attribution changes later. Record what can be seen, how it was seen and why the example was trusted or limited.

Identity

unique research number
object title, maker or publisher
claimed edition or variant
owner, institution, accession or catalogue number

Source and access

physical examination, institutional record, auction, listing or private submission
observation date and stable archive reference
examined in hand, through glass, by high-resolution image, limited image or description only

Physical and historical data

dimensions, weight, materials, construction and markings
contents, condition, provenance and restoration history
previous sales, ownership labels and documentation gaps

Images and rights

overall views and diagnostic details
scale, image source and photographer
usage rights where relevant

Confidence

identity confidence
date and chronology confidence
completeness and originality confidence
provenance and condition-interpretation confidence

Independence

ownership and discovery source
seller or consignment relationship
same object reappearing
suspected batch or copied-attribution relationship

A compact comparison matrix

Field	Purpose
Comparator ID	Keeps examples and repeat appearances distinct
Evidence basis	Shows why the attribution is believed
Region / date	Controls geography and chronology
Materials / dimensions	Tests construction and format
Diagnostic traits	Records presence, absence or uncertainty
Condition / completeness	Prevents damage or missing parts being treated as variation
Examination mode	Separates physical inspection from image-only evidence
Independence	Flags same-object, same-batch and copied-source relationships
Confidence	Stops uncertain examples carrying full evidential weight

Record absence carefully: present, absent, obscured, not photographed, not examined, uncertain and not applicable are different states. “Not visible” must not be converted into “absent.”

Weight examples rather than counting them equally

Three high-quality comparisons may support a stronger conclusion than twenty anonymous listings. Weight should reflect at least identity confidence, observation quality, provenance quality, condition interpretability and independence.

Identity

Observation

Provenance

Condition

Independence

Write conclusions that preserve uncertainty

The strength of the wording should track the strength and limits of the comparison evidence. This keeps provisional findings useful without turning them prematurely into catalogue facts.

Strong support

All eight securely identified revised UK examples carry the same printer code, while none of the six documented earlier examples does.

Moderate support

The available evidence suggests that the code was introduced during the revised issue, although only two transitional examples have been examined.

Limited support

This feature may indicate the later issue, but the conclusion currently rests on three image-only comparisons.

Inconclusive

The observed difference can also be explained by fading and photographic variation; the current examples do not support a variant classification.

Myth versus reality

Myth: the best-looking example is the best reference.

Reality: an altered pristine example may conceal more production evidence than a worn, untouched one.

Myth: many matching listings equal strong corroboration.

Reality: listings may repeat one object, one batch or one inherited description. Independence matters more than raw count.

Myth: it does not match the known fake, so it is genuine.

Reality: it only lacks the recognised features of that counterfeit type. Other false constructions may exist.

Myth: sealed means untouched and complete.

Reality: a seal is an observed condition claim. Its originality, internal contents and any pre-sealing substitution still require evidence.

Domain boundaries

Comparison informs neighbouring decisions but does not replace them

Authentication: comparator selection contributes to an authenticity assessment, but a final judgement may also require provenance, specialist examination and scientific analysis.

Grading and condition: comparison helps distinguish normal variation, ageing and intervention, but it is not itself a formal grade.

Valuation: market comparables must be selected separately from identity comparators and interpreted within the relevant sale context.

Preservation and restoration: research may reveal treatment or risk, but collectors should not clean, dismantle or test an object merely to improve comparability.

When to escalate

Specialist involvement is warranted when the comparison problem exceeds ordinary visual and documentary research, or when the cost of being wrong is high.

The attribution or classification has major financial consequences.
Scientific testing, ultraviolet examination or invasive access is being considered.
The object is exceptionally rare and the available comparison population is very small.
Sophisticated counterfeiting, extensive restoration or concealed replacement is suspected.
Provenance gaps, legal title or cultural-property questions affect the research.
The conclusion depends on highly specialised workshop, factory, printing or material knowledge.

A specialist opinion should still disclose the objects and records examined, the comparison set used, the limits of access, the reasoning and the degree of certainty. “An expert says so” is evidence of opinion, not a substitute for documenting its basis.

Key takeaways

Choose comparison examples for the precise question being tested, not for general resemblance.
The identity, condition, provenance and examination quality of each comparator determine how much weight it can carry.
Use groups containing representative, diagnostic, boundary, negative and anomalous examples rather than relying on one ideal-looking object.
Control chronology, geography, manufacturing source, materials, dimensions, condition, completeness and restoration before interpreting difference.
Record observations, absences, uncertainty, source dependence and same-object reappearances before writing the conclusion.
A defensible conclusion explains what each comparison proves, what it cannot prove and how the result would change if the comparison set changed.

Continue learning

Back to Comparative Analysis

Return to the comparative analysis overview and the wider sequence of collector comparison methods.

→

Comparative Analysis

Explore how collectors use comparison to recognise patterns, test claims and qualify uncertainty.

→

Like-for-Like Comparison

Continue by examining whether two examples are truly comparable or only superficially similar.

→