What Information Can Investigators Get From a Blood Sample Collected at a Crime Scene?
Separating what a bloodstain contains from what investigators infer about how it got there
Executive Summary
A dried bloodstain the size of a coin can, in principle, be interrogated at many different molecular levels: as intact and degraded cells, as haemoglobin and its breakdown products, as nuclear DNA, as RNA and methylated DNA, as a protein and peptide inventory, and as a physical shape shaped by fluid dynamics. Each level is read by a different instrument and answers a different question. Established forensic biology can, under favourable conditions, tell an investigator that a stain is blood, that it is human, and whose genetic profile it carries. Past that point, the science becomes markedly less certain and far more contested. Bloodstain age estimation, biological-characteristic prediction from DNA, and the reconstruction of events from spatter patterns remain partly experimental, statistically bounded, or actively disputed within the forensic science community itself. This review organises that uneven landscape into a layered framework and holds one distinction constant throughout: what the sample contains is not the same claim as what happened.
Key Findings
- Presumptive blood tests (Kastle-Meyer, Hemastix, luminol) are highly sensitive — some down to dilutions of 1 in 100,000 — but only moderately specific, producing documented false positives from plant peroxidases, oxidising metals, and household bleach.
- DNA profiling reliably associates a stain with a source at the sub-source or source level of the hierarchy of propositions, but mixture interpretation using probabilistic genotyping software rests on a set of contestable statistical assumptions that require rigorous, casework-realistic validation (Thompson, 2023).
- A comprehensive 2025 review of bloodstain age estimation by ultraviolet-visible spectroscopy concludes the field still lacks a single environmentally robust, court-ready method, despite roughly three centuries of research effort (Bergmann et al., 2025).
- Large "black box" reliability studies of bloodstain pattern analysis found meaningful rates of analyst disagreement and error even among experienced practitioners (Hicklin et al., 2021; National Institute of Justice, 2022).
- Forensic DNA phenotyping predicts some pigmentation traits with useful accuracy, but ancestry and full-appearance prediction carry population dependence, statistical uncertainty, and documented risks of discriminatory misuse (Schneider et al., 2019).
- Documented cognitive bias from task-irrelevant contextual information measurably shifts forensic conclusions, including DNA mixture calls and bloodstain pattern classifications (Dror, 2018; Dror & Hampikian, 2011).
- Under India's Bharatiya Sakshya Adhiniyam, 2023, and Section 176(3) of the Bharatiya Nagarik Suraksha Sanhita, 2023, forensic attendance at serious crime scenes is now a statutory obligation — but India-specific validation data for several of the techniques discussed in this article remain scarce, a gap this review does not paper over.
Introduction: The Information Hidden in a Blood Sample
A coin-sized bloodstain, dried onto a wooden floor, may contain intact and ruptured red cells, denaturing haemoglobin, degrading messenger RNA, methylated DNA carrying an epigenetic age signal, a peptide inventory distinct from every other body fluid, and nuclear DNA capable of generating a full genetic profile. It also has a shape, an edge pattern, a distance from the nearest wall, and a position relative to a doorway. In sheer molecular and physical terms, this is an extraordinarily information-dense object.
And yet the same stain may still be unable to answer the questions an investigator cares about most. Who was injured? When, precisely, did the blood leave the body? How did it arrive on that floor — dripping, transferred by a shoe, cast off from a swinging object? Was the person who contributed the DNA present when the injury occurred, or does the stain represent an entirely unrelated, earlier event?
This is the paradox at the centre of blood evidence: enormous molecular information density does not translate cleanly into investigative certainty. This article asks, deliberately, how much information can actually be extracted from a blood sample — and where does the scientific information end and forensic interpretation begin? It is not a basic-level explainer of "what blood evidence can do." It is an attempt to map the boundary between detection, analysis, association, and reconstruction, and to show why conflating those steps is one of the most consequential errors in forensic reasoning.
Blood Is More Than DNA Evidence: The Forensic Information Layers of Blood
Popular understanding collapses "blood evidence" into "DNA evidence." Forensic biology does not. Whole blood is a suspension of red cells, white cells, and platelets in plasma, itself a solution of proteins, electrolytes, metabolites, and (in the case of a living or recently deceased contributor) whatever exogenous substances were circulating at the time. Different analytical techniques interrogate entirely different components of this mixture, and each component degrades, persists, and informs on a different timescale.
This review proposes a working framework — the Forensic Information Layers of Blood — solely to organise the discussion that follows. It is not an established forensic standard; it is a conceptual scaffold developed for this article based on the scientific literature reviewed throughout.
The Nine Layers
| Layer | Core Question | Typical Method |
|---|---|---|
| 1. Biological Identification | Is this substance biological material, and specifically blood? | Presumptive colorimetric/chemiluminescent tests |
| 2. Species and Biological Origin | Is it human blood, or from another species? | Immunological confirmatory tests; species-specific DNA markers |
| 3. Individualisation and Association | Can it be associated with a specific person or contributor population? | STR profiling; probabilistic genotyping |
| 4. Biological Characteristics | What externally visible or biological traits might the contributor have? | Forensic DNA phenotyping (sex, ancestry, pigmentation) |
| 5. Physiological / Chemical State | Was a substance present, or was a physiological state detectable? | Toxicology; biomarker/protein assays |
| 6. Temporal Information | Can degradation patterns suggest how old the stain is? | Spectroscopy; RNA/protein degradation modelling |
| 7. Spatial and Pattern Information | What does the stain's shape and location suggest? | Bloodstain pattern analysis; scene documentation |
| 8. Event-Level Inference | What hypotheses about the event can be evaluated with all evidence combined? | Integrated case evaluation; Bayesian reasoning |
| 9. The Epistemic Boundary | What can the evidence never determine, regardless of technique? | Not a method — a limit |
The remainder of this article works through these layers roughly in order, pausing at each to separate what is analytically established from what remains investigative or experimental — and, throughout, to separate the analytical result itself from the inference an investigator might draw from it.
Observation, Result, and Inference: A Recurring Ladder
Before proceeding, it is worth naming the interpretive chain this article returns to repeatedly. A forensic conclusion typically moves through several stages: an observation (a reagent changes colour), an analytical result (a DNA profile is generated), a source-level inference (the profile is associated with a named individual), an activity-level inference (that individual's blood was deposited by a particular action), an event reconstruction (a sequence of events is proposed), and finally an investigative narrative (a story about what happened and why). Forensic geneticists have formalised the middle portion of this chain as the hierarchy of propositions — offence, activity, source, and sub-source levels — precisely because each upward step requires additional information beyond the biological result itself, and each step increases the assumptions in play (Cook et al., 1998; Evett et al., 2000; Evett et al., 2002; Gittelson et al., 2016). A forensic scientist's finding, properly reported, belongs at the source or sub-source level; the higher-level propositions about activity and offence are, by design, for the court and the trier of fact.
Figure: The observation-to-narrative chain, showing where a strictly biological finding ends and where additional inference, context, and assumptions begin.
Establishing Whether a Suspected Stain Is Blood
Before any biological or genetic question can be asked, an examiner must decide whether a red-brown stain is blood at all. This is done in two stages: presumptive testing, which is fast, cheap, and sensitive but not conclusive; and confirmatory testing, which is slower and more specific.
The most common presumptive reagents — phenolphthalein (the Kastle-Meyer test), tetramethylbenzidine, leucomalachite green, and luminol — all exploit the peroxidase-like activity of haemoglobin, producing a colour change or, in luminol's case, chemiluminescence, in the presence of hydrogen peroxide. Sensitivity is generally excellent: comparative validation work has reported luminol and Kastle-Meyer detecting blood diluted to roughly 1 in 100,000 and 1 in 16,384 respectively. Specificity is the weaker property. Kastle-Meyer has produced false positives from potato and horseradish peroxidase; Hemastix has cross-reacted with tomato, rust, bleach, and avian uric acid; and luminol is well known to react with copper salts, certain alloys, and household bleaching agents. One controlled specificity study found that bleaching agents alone accounted for a large share of false positive reactions across several presumptive reagents used routinely in casework. Because presumptive tests can also be chemically destructive to downstream DNA typing — Kastle-Meyer has been shown to inhibit amplification at low dilutions in some protocols — laboratories increasingly weigh sensitivity, specificity, and DNA compatibility together rather than defaulting to a single reagent.
Confirmatory methods narrow this uncertainty. Microscopic identification of red blood cell morphology, and the classical Takayama (haemochromogen) and Teichmann (haematin) crystal tests, remain specific for haemoglobin-derived material, though they are comparatively insensitive and technically exacting. Modern casework has largely shifted confirmatory and species work onto immunological and DNA-based platforms, discussed next.
Figure: A presumptive positive is provisional; only a confirmatory test (or an equivalent DNA-based confirmation) supports a definitive identification of blood.
Human Versus Non-Human Biological Origin
A confirmed bloodstain still carries an open question: is it human? Historically this was answered by the precipitin reaction — an antiserum raised against human serum proteins forms a visible precipitate when it meets human blood, a principle exploited since the early twentieth century in techniques such as the Ouchterlony double-immunodiffusion test. Enzyme-linked immunosorbent assays (ELISA) later improved sensitivity and, in several validated protocols, showed no cross-reaction against panels of livestock species such as pig, sheep, cattle, goat, horse, and rabbit immunoglobulin. Cross-reactivity risk rises specifically among closely related species — distinguishing human blood from that of other primates has historically been the harder discrimination problem for antibody-based methods, because shared evolutionary ancestry means shared antigenic sites.
In current practice, human-origin confirmation is frequently folded into the same molecular workflow used for DNA profiling: species-specific primers, or the observation that a human STR or amelogenin assay simply fails to amplify non-human template, can substitute for or corroborate a separate immunological test. Infrared spectroscopic methods that distinguish human from animal blood by their vibrational signatures have also been demonstrated experimentally, though these remain less standardised in casework than either serology or DNA-based species testing (Mistek-Morabito & Lednev, 2020).
The central question at this layer is a threshold question, not a technical one: what level of certainty is required before an investigator moves from "possibly blood" to "confirmed human biological evidence" — and does the laboratory's reporting language honestly reflect that threshold, or does it quietly overstate it?
What DNA Can and Cannot Tell Investigators
Short tandem repeat (STR) profiling is the analytical backbone of modern forensic biology, and for good reason: it is highly discriminating, well validated, and — for single-source, sufficient-quantity samples — capable of producing a profile whose random-match probability can be astronomically small. But crime-scene blood samples are frequently not single-source or sufficient-quantity. They are commonly degraded, low in template DNA, or mixed with material from more than one contributor, and each of those conditions changes what a DNA result can honestly claim to show.
Detection — the observation that DNA is present and of a given quantity and quality — is distinct from identification, the generation of a genotype at each tested locus. Identification is distinct again from association, the statistical comparison of that genotype against a reference sample or database entry, typically expressed as a likelihood ratio. And association is distinct from interpretation, the process by which an analyst decides how confidently, and at what proposition level, that statistical result should be reported.
Mixture interpretation is where this chain becomes genuinely contested. Complex mixtures — those with more than one contributor, especially at low template levels — historically relied on binary allele-counting methods that struggled with peak height variation, allele drop-out, and drop-in. Probabilistic genotyping software (such as STRmix and TrueAllele) now models these stochastic effects statistically, producing continuous likelihood-ratio outputs rather than binary match/non-match calls (Gittelson et al., 2016). This is a genuine scientific advance, but it is not an uncontested one: a 2023 case study directly comparing two widely used probabilistic genotyping platforms on the same low-template evidence found materially different likelihood-ratio outputs and concluded that such analysis "rests on a lattice of contestable assumptions," calling for clearer validation and more transparent reporting of those assumptions in court testimony (Thompson, 2023). A published commentary on that same case study pushed back on some of its framing while agreeing that rigorous, casework-realistic validation of probabilistic genotyping software remains essential (Kalafut et al., 2023). The disagreement between these two peer-reviewed pieces is itself instructive: even DNA mixture interpretation, often treated by the public as the "gold standard" of forensic science, contains genuine, ongoing scientific debate about how confidently its outputs should be stated.
None of this diminishes the core reliability of single-source STR profiling under good conditions. It does mean that a DNA "match" reported from a crime-scene bloodstain should be read as a specific, bounded claim — a genotype was generated and associated with a source at a stated statistical strength, under stated assumptions — not as a general assertion that "the defendant's blood was definitely at the scene." That gap between the analytical result and its plain-language restatement is precisely where source-level and activity-level propositions diverge (Evett et al., 2002).
Beyond DNA: The Molecular Information Inside Blood
Nuclear DNA sequence is only one of several molecular signals blood carries. Because RNA transcription and epigenetic modification are tissue-specific and time-dependent in ways raw DNA sequence is not, three adjacent research fields have grown rapidly: forensic RNA typing, DNA methylation analysis, and protein/peptide profiling. Each is at a different point on the path from laboratory discovery to courtroom-ready method.
| Technology | Current Scientific Status | Validation Maturity | Potential Forensic Information | Major Limitation |
|---|---|---|---|---|
| mRNA body-fluid typing | Validated in specialised contexts | Multi-lab collaborative exercises completed | Distinguishes blood from saliva, semen, vaginal fluid, menstrual blood | RNA degrades faster than DNA; tissue specificity imperfect |
| DNA methylation (epigenetic clocks) | Validated in specialised contexts | Multiple independent models (e.g., ELOVL2 locus); accuracy of a few years reported in research cohorts | Chronological age estimation; some body-fluid discrimination | Bisulfite conversion reduces template; environmental/lifestyle confounds (e.g., smoking) shift methylation |
| MicroRNA / circular RNA age markers | Emerging research | Single/limited-cohort proof-of-concept studies | Alternative age-estimation marker set, potentially more stable than mRNA | Small sample sizes; cross-cohort replication still limited |
| Forensic proteomics (mass spectrometry) | Validated in specialised contexts / Emerging | Peptide biomarker panels published and cross-validated for several body fluids; broader adoption limited | Body-fluid identification (including from DNA-extraction waste), some post-mortem-interval and biological-age signal | High equipment cost; sample stability and standardisation still developing (Chhikara et al., 2026) |
| Nanopore / long-read multi-marker sequencing | Emerging research | Exploratory, low-DNA-input studies | Single-assay combined age and body-fluid signal | Low read-depth samples show measurement instability; correction models still proof-of-concept |
| Extracellular vesicle multi-omics | Exploratory research | Almost entirely clinical/biomedical, not forensic, at present | Speculative future body-fluid or disease-state signal | No meaningful forensic-specific validation literature yet identified for crime-scene bloodstains |
The methylation-based "epigenetic clock" concept deserves particular attention because it is often the most overstated of these techniques in public discussion. Research has shown that blood-based DNA methylation at a small number of CpG sites correlates with chronological age closely enough to be modelled statistically (Weidner et al., 2014), and follow-up work has targeted specific loci — the ELOVL2 gene among the most consistently reported — for age-prediction panels (Jung et al., 2019). A broader multi-marker study combining mRNA, DNA methylation, DNA rearrangement, and telomere length in blood reported that combined models improved on any single marker type alone (Zubakov et al., 2016). More recent work applying microRNA and circular RNA markers to the same age-estimation problem represents a genuinely newer, less-validated branch of this research (Wang et al., 2022). None of this amounts to a court-ready "read the exact age off the bloodstain" capability; it amounts to a statistically bounded estimate with a margin of error, built on population-level training data whose applicability to an unknown individual from an unknown population always carries some uncertainty.
Forensic proteomics occupies a similar middle ground. Mass spectrometry-based approaches have identified tissue-specific peptide biomarkers capable of discriminating blood from saliva, semen, vaginal fluid, and — notably — distinguishing menstrual blood from peripheral blood, a discrimination immunoassays historically found difficult. A 2026 review in the Journal of Proteome Research catalogued growing use of proteomic and machine-learning methods for body-fluid classification and post-mortem-interval estimation, while explicitly flagging that "wide acceptance of forensic proteomics remains problematic due to intricate sample stability, high equipment costs, and strict legal standards" (Chhikara et al., 2026). That is an honest summary of where the field stands: real analytical capability, immature operational infrastructure.
Can Blood Reveal Biological Characteristics?
Forensic DNA phenotyping (FDP) — predicting externally visible characteristics such as eye, hair, and skin colour, biogeographic ancestry, and age directly from crime-scene DNA — exists specifically to generate investigative leads when no database match is available (Kayser, 2015). It is worth being precise about what this technology does and does not achieve, because public reporting often blurs the two.
Pigmentation prediction is the most scientifically mature branch. Validated SNP-based tools such as the IrisPlex and HIrisPlex-S systems can categorise eye, hair, and skin colour with reported accuracy — measured as area-under-the-curve values — ranging roughly from the low 0.6s to the high 0.9s depending on the trait and colour category, meaning performance is strong for some categories (e.g., blue versus brown eyes) and considerably weaker for intermediate categories (Schneider et al., 2019). Sex determination from blood — via amelogenin or Y-chromosome markers — is comparatively simple and highly reliable in good-quality samples, and is treated as a routine part of standard STR profiling rather than as a separate "phenotyping" exercise.
Ancestry inference and full facial-appearance prediction are considerably less settled. Biogeographic ancestry panels can assign continental-level ancestry with reasonable confidence but struggle with admixed populations and finer-grained regional distinctions — a limitation with direct relevance in a demographically diverse country like India, where continental-ancestry categories map poorly onto local population structure. Legal and ethics scholarship has additionally raised sustained concern that ancestry and appearance prediction risk contributing to discriminatory profiling and stigma if presented to investigators or juries without careful statistical framing (see the extended discussion in law-review literature on regulating genetic appearance estimation). Age prediction from DNA methylation, discussed in the previous section, is often grouped with FDP for the same reason: it produces a population-based estimate with a margin of error, not a determinate fact.
Chemical and Toxicological Information
Blood is also a chemical record, and forensic toxicology treats it as such — but the record's readability depends enormously on whether the blood was drawn from a living person, collected as fresh post-mortem blood, or recovered as an aged, dried crime-scene stain.
Dried blood spot (DBS) methodology, adapted from clinical newborn-screening practice, has been validated for a range of forensic toxicology applications: multiple studies report drugs of abuse and psychotropic medications remaining detectable and quantifiable in dried spots for periods ranging from roughly one to eight months depending on storage temperature, with colder storage consistently improving analyte stability. One post-mortem DBS validation study screened dozens of antidepressant and antipsychotic compounds and successfully quantified several in real casework samples, while explicitly monitoring degradation over a three-month period as part of the validation itself (Moretti et al., 2019). This is a genuine and growing capability — but it answers a narrower question than it might appear to: it can show that a compound was present at a measurable concentration at the time of testing, adjusted for known degradation, not necessarily the concentration circulating in the body at the moment of the event under investigation.
That caveat becomes critical once post-mortem physiology enters the picture. Drug concentrations in blood collected after death are subject to post-mortem redistribution — the movement of drugs between tissue compartments as cell membranes break down after death, a phenomenon toxicologists have described, in a frequently cited phrase, as a "toxicological nightmare" precisely because it can inflate or deflate measured concentrations depending on which vessel the sample was drawn from and how much time has elapsed (Pounder & Jones, 1990). A measured drug concentration in post-mortem blood is therefore never automatically equivalent to the concentration present at the moment of death, let alone proof of impairment or intent at an earlier point in time.
The central question this section keeps returning to: does detecting a chemical compound in blood automatically reveal when, or under what circumstances, exposure occurred? The answer, consistently, is no — detection establishes presence and, with careful quantitative work, a bounded concentration estimate; timing and circumstance require additional contextual and pharmacokinetic reasoning that the chemical result alone cannot supply.
Can Blood Reveal Disease or Physiological State?
Clinical medicine routinely uses blood biomarkers — troponins, inflammatory markers, metabolic panels — to characterise disease and physiological state in living patients under controlled sampling conditions. Forensic application of the same logic to an unidentified crime-scene stain is far more constrained, for reasons that are more about context than chemistry. A biomarker validated in a clinical cohort, drawn under known timing and known patient history, does not automatically transfer its diagnostic meaning to a dried, environmentally exposed, source-unknown stain of uncertain age. Protein degradation, environmental exposure, and the simple absence of a clinical baseline for comparison all weaken the inferential chain between "this biomarker is elevated" and "this person had this condition."
This review therefore treats disease or physiological-state inference from crime-scene blood as, at most, an exploratory research direction rather than an established forensic capability, and takes care not to imply that any biomarker discussed elsewhere in this article — age-related methylation markers, proteomic panels, or toxicological findings — amounts to a medical diagnosis of an unidentified individual. Where clinical biomarker research is cited in this article (for example, in the discussion of multi-omics and extracellular vesicles), it should be read as background on the underlying biology, not as evidence of forensic-readiness.
The Challenge of Bloodstain Age Estimation
If there is one question the public most consistently overestimates forensic science's ability to answer, it may be this one: how old is this bloodstain? A 2025 comprehensive review of ultraviolet-visible spectroscopic approaches to bloodstain age estimation (BAE) — tracing research back roughly three centuries — makes the state of the field explicit: UV-Vis spectroscopy is sensitive to temperature, humidity, and light exposure, and provides comparatively limited molecular specificity, making it "challenging to ensure consistent accuracy in diverse settings" (Bergmann et al., 2025).
The underlying biology explains why. As a bloodstain dries and ages, oxyhaemoglobin oxidises progressively to methaemoglobin and eventually to hemichrome, producing measurable shifts in absorbance and reflectance spectra (Hanson & Ballantyne, 2010; Bremmer et al., 2011). In parallel, RNA degrades, serum proteins denature, and — on a slower and more DNA-degradation-dependent timescale — nuclear DNA itself fragments. Multiple analytical platforms have been applied to track these changes: UV-Vis and near-infrared reflectance spectroscopy, Raman spectroscopy, attenuated total reflectance Fourier-transform infrared (ATR-FTIR) spectroscopy, high-performance liquid chromatography analysis of haemoglobin derivatives (Andrasko, 1997), atomic-force/force spectroscopy of the physical properties of ageing blood (Strasser et al., 2007), aspartic acid racemization kinetics (Arany & Ohtani, 2011), and even smartphone-camera colorimetric analysis intended for field-deployable screening (Thanakiatkrai et al., 2013).
One of the more methodologically rigorous studies in this space applied ATR-FTIR spectroscopy combined with chemometric modelling to bloodstains aged under simulated indoor and outdoor crime-scene conditions for up to 107 days, reporting good statistical performance for distinguishing fresh from older stains and for regression-based age prediction within that window (Lin et al., 2017). That is a genuinely strong result — but it was achieved under simulated, controlled conditions with a specific substrate and environment, which is precisely the caveat the wider literature keeps repeating: laboratory proof-of-concept performance under controlled variables does not automatically transfer to the far more variable substrates, temperatures, humidity levels, and light exposures of a real, uncontrolled crime scene.
Information From Location and Context
Everything discussed so far concerns the biological sample itself. But a bloodstain also has a location: which surface it is on, how far it is from a doorway or a weapon, what other stains surround it, and what object it may have contacted before or after deposition. This contextual information is genuinely useful for building investigative hypotheses and, later, for reconstruction — but it is important to be precise about where it comes from. It is not information contained within the blood sample's biology; it is information about the scene that happens to be associated with the sample's position. Treating spatial context as though it were a biological property of the blood itself is a subtle but consequential category error, because it can smuggle scene-level assumptions into what should be a narrowly bounded biological or genetic conclusion.
Bloodstain Pattern Analysis: Information From the Pattern, Not the Sample
Bloodstain pattern analysis (BPA) deserves to be treated as conceptually distinct from the biological analysis of blood discussed in earlier sections, because it asks a fundamentally different kind of question: not "whose blood is this," but "what physical mechanism produced this shape, and from what direction and distance did it originate." BPA's theoretical foundation — fluid dynamics, drop trajectory physics, and impact-angle geometry — is genuinely scientific. Its practical application in casework has, however, come under sustained and serious scrutiny over the past fifteen years.
The 2009 National Research Council report on forensic science in the United States singled out bloodstain pattern interpretation for particular concern, describing the uncertainties involved as "enormous" and characterising many analysts' opinions as more subjective than scientific (National Research Council, 2009). Large-scale empirical "black box" studies conducted since have given that concern quantitative shape. One reliability study assessing pattern classification found substantially higher inter-rater agreement on rigid, non-absorbent surfaces than on fabric, where classification proved considerably less consistent (Taylor et al., 2016). A subsequent large error-rate study, funded by the National Institute of Justice and involving 75 practising analysts examining stains produced under known conditions, produced what the Institute itself described as one of the first large error-rate studies of the discipline — finding that a meaningful share of incorrect responses were not caught even when a second analyst reviewed the same evidence (Hicklin et al., 2021; National Institute of Justice, 2022).
A related and equally important finding from this same body of research is that bloodstain-pattern analysts, in practice, often use contextual case information when making pattern classifications, which measurably influences their conclusions — meaning the boundary between "reading a pattern" and "reconstructing a scene using outside information" is, in practice, frequently blurred (National Institute of Justice, 2022). More recent published critiques go further, characterising BPA's validity, reliability, and vulnerability to cognitive bias and error as areas of serious ongoing concern even while acknowledging that its underlying physical principles are legitimate, and recommending reforms such as refined classification standards, bias-mitigation protocols like Linear Sequential Unmasking, and increased use of 3D scanning and machine-learning-assisted feature extraction to reduce subjectivity going forward.
None of this means BPA is scientifically worthless — its physical premises are sound, and reform efforts are active and specific. It does mean that a pattern classification, on its own, should not be treated as though it carries the same evidentiary weight as a validated laboratory measurement, and that BPA-based reconstruction claims deserve at least as much scrutiny for analyst-to-analyst reliability as for the underlying physics.
Transfer, Persistence and Alternative Explanations
Underlying every claim in this article is a principle articulated more than a century ago by Edmond Locard: contact between two surfaces produces an exchange of material in both directions. This is usually shortened to "every contact leaves a trace" — a memorable but incomplete summary, because a trace only helps an investigation if it was transferred in the first place, survived the interval before recovery, and was actually collected by the method used. Miss any one of those three steps — transfer, persistence, recovery — and the trace is absent even though contact genuinely occurred.
DNA transfer research over the past two decades has made this principle considerably more complicated than its slogan suggests. A comprehensive review of DNA transfer, persistence, prevalence, and recovery (TPPR) catalogued the many variables that affect whether, and how much, DNA moves from a contributor to a surface, and cautioned that terminology implying a specific source or activity should not be used unless supporting evidence is actually available (van Oorschot et al., 2019). Controlled studies have specifically demonstrated secondary DNA transfer — material moving from person A to an object via an intermediary person B who never touched the object directly — under experimental conditions (Goray et al., 2010), and laboratory studies have shown that ordinary examination tools such as scissors, forceps, and gloves can themselves transfer detectable DNA between exhibits during routine handling (Szkuta et al., 2015).
These findings do not undermine the value of blood or DNA evidence generally. They do sharpen the central question this section poses: does detecting blood, or a DNA profile within it, establish the activity that caused its presence? On the evidence reviewed here, the answer is that it does not, automatically — transfer mechanism, persistence over time, and the possibility of indirect or secondary deposition must each be separately considered before a biological result is safely converted into an activity-level claim (de Zoete et al., 2016).
What Happens When Blood Evidence Is Combined With Other Evidence?
Blood evidence rarely stands alone in a real investigation; it sits alongside fingerprints, digital records, CCTV, witness accounts, toxicology, and pathology findings. Combining evidence streams can genuinely increase the overall strength of a case — but it is worth being precise about the mechanism. Integration does not make the blood sample itself contain more information than it did in isolation. What changes is the surrounding context available to interpret the same biological result: a DNA association that is ambiguous between an innocent-contact and a crime-related explanation may become far less ambiguous once combined with, say, digital location data placing the same person at the scene at a relevant time, or entirely more ambiguous if a witness account establishes an innocent explanation for contact. The value of combination is contextual and case-specific, not an automatic multiplier — "more evidence" does not mechanically equal "more certainty" unless the additional evidence actually bears on the specific proposition in dispute, a principle closely related to the hierarchy-of-propositions framework discussed earlier (Evett et al., 2002).
The Information-to-Inference Gap: The Blood Evidence Inference Ladder
Drawing together the sections above, this review proposes a second organising framework — the Blood Evidence Inference Ladder — to make explicit how far a single biological finding can be legitimately extended before additional evidence and additional assumptions are required. Like the Forensic Information Layers presented earlier, this is a conceptual model developed for this article, not an established professional standard.
| Level | Claim | What Is Actually Known | Assumptions Required to Go Further |
|---|---|---|---|
| 1 | A substance is detected | A reagent reacted; a spot exists | That the reaction is not a false positive |
| 2 | The substance is identified as blood | Confirmatory test/microscopy supports haemoglobin-derived origin | Adequate confirmatory testing was actually performed |
| 3 | Biological characteristics are identified | Species/sex/phenotype markers produce a result | Marker panels are validated for the relevant population |
| 4 | A DNA profile / biological association is generated | A genotype exists and has been statistically compared | Mixture assumptions (number of contributors, drop-out/drop-in rates) are sound |
| 5 | Evidence associated with an individual/contributor hypothesis | A likelihood ratio or match statistic at source/sub-source level | The reference population and propositions are correctly framed |
| 6 | Inference about how blood was deposited | Nothing directly — this is now a transfer/persistence hypothesis | Transfer, persistence, and recovery dynamics for this scenario are understood |
| 7 | An activity or event is proposed | Nothing directly from the sample — requires case context | Alternative explanations for the activity have been considered and excluded |
| 8 | A historical narrative is reconstructed | Nothing directly from the sample — an integrative, investigative judgment | All prior levels hold, and the narrative is not one of several equally consistent alternatives |
Figure: The Blood Evidence Inference Ladder — each rising level narrows the sample's direct contribution and widens the assumptions required to sustain the claim.
What a Blood Sample Cannot Tell Investigators By Itself
Consolidating the caveats raised throughout this article: under ordinary forensic interpretation, a blood sample alone generally cannot establish the complete event that caused bleeding, a person's intention or motive, guilt, an exact deposition time, an exact sequence of events, why a person was present at a location, or whether a person was present at the moment a crime occurred. Each of these questions sits at the activity or offence level of the hierarchy of propositions — levels the biological evidence, by itself, was never designed to answer (Evett et al., 2002). Establishing them legitimately requires additional evidence, transparent reasoning about alternative explanations, and — for the highest-level questions of guilt and intent — a legal process that goes well beyond forensic science's remit.
Information Potential Versus Forensic Reliability
The intellectual thread running through this entire article can be stated as a short chain of non-equivalences:
Information can be extracted does not necessarily mean information can be reliably interpreted.
Information can be interpreted does not necessarily mean information can establish a historical event.
This chain explains why methods discussed in this article cluster at very different points on the reliability spectrum despite all originating from the same biological source material. Species identification and single-source STR profiling sit close to the "reliably extracted and interpreted" end. Bloodstain age estimation and bloodstain pattern reconstruction sit further along, where extraction is often possible but interpretation carries substantial, well-documented uncertainty. Disease-state inference from an unidentified stain sits close to the "information may exist, but cannot presently be reliably extracted for forensic purposes" end. Recognising which point on this spectrum a given technique occupies — rather than treating "it's science" as a single undifferentiated stamp of reliability — is arguably the most transferable skill this review can offer a forensic practitioner, student, or legal professional.
Emerging Molecular Technologies and the Future of Blood as a Forensic Information Archive
Several research directions may extend what can be responsibly extracted from crime-scene blood in the coming years, though each currently carries a meaningful validation gap. Multi-marker approaches that combine methylation, RNA, and protein signals in a single assay — rather than testing each in isolation — have shown promise for improving age-estimation accuracy over any single marker type (Zubakov et al., 2016). Machine-learning-assisted spectroscopic modelling is being applied increasingly to bloodstain age estimation, with the aim of better handling the substrate and environmental variability that has historically limited single-technique approaches (Bergmann et al., 2025; Lin et al., 2017). Long-read and nanopore sequencing platforms are being explored as a way to recover combined age and body-fluid signals from very small DNA quantities in a single sequencing run, though early studies report that low sequencing depth introduces measurement instability that correction models can only partially address. Mass spectrometry-based proteomics continues to expand the list of validated peptide biomarkers for body-fluid and mixture discrimination (Chhikara et al., 2026; Macfarlane et al., 2026).
Extracellular vesicles and broader multi-omics integration are sometimes proposed as the "next frontier" for forensic biology, largely by analogy with rapid progress in clinical and cancer biomarker research. This article treats that possibility cautiously: the extracellular-vesicle multi-omics literature reviewed for this piece is, at present, almost entirely clinical and biomedical rather than forensic, and no meaningful forensic-specific validation studies applying these methods to crime-scene bloodstains were identified during research for this article. Readers should treat this area as a plausible future direction supported by adjacent biomedical science, not as an existing forensic capability.
Across all of these directions, the same three questions from the scientific-promise framework recur: does the underlying biology support the claimed signal; has it been validated across realistic, variable crime-scene conditions rather than controlled laboratory conditions alone; and is the result reproducible across independent laboratories and population groups? Operational readiness, in every case surveyed here, lags meaningfully behind laboratory proof-of-concept.
The India Context: Statutory Mandate, Uneven Validation
India's criminal procedure has recently undergone its most significant forensic-evidence reform in decades. The Bharatiya Sakshya Adhiniyam, 2023, replacing the Indian Evidence Act, 1872, elevates electronic and certain scientific records within the evidentiary framework, while Section 176(3) of the Bharatiya Nagarik Suraksha Sanhita, 2023 — in force since 1 July 2024 — makes it a statutory obligation for a forensic expert to physically visit the crime scene, collect evidence, and videograph the process in offences punishable by seven years' imprisonment or more. Both changes place considerably more weight on scientific evidence, including blood evidence, than earlier procedural law did.
That statutory weight has not been matched, in the material surveyed for this article, by an equivalent expansion of India-specific validation research for several of the techniques discussed above. Bloodstain age estimation, forensic DNA phenotyping accuracy across India's ethnically and regionally diverse population, and forensic proteomic panels have each been validated primarily on cohorts and substrates studied elsewhere — chiefly in East Asian, European, and North American laboratories. This is not a criticism unique to India; it reflects a broader pattern in which population-specific and substrate-specific validation lags the publication of a technique's initial proof-of-concept almost everywhere. But it is a gap worth stating honestly rather than assuming away: a forensic expert applying, say, an ancestry-inference panel or a spectroscopic age-estimation model to Indian casework should recognise that the underlying training and validation data may not fully represent the population or environmental conditions of the case at hand, and should communicate that limitation candidly if called to testify.
Suggested Infographic Concept
The Information Hidden in a Crime-Scene Blood Sample
The figure below visualises the progression from raw biological material to investigative narrative, with colour coding separating directly measured information from inferred information and from information that cannot be established from the sample alone.
Figure: Conceptual progression of forensic information layers in crime-scene blood evidence, from directly measured biological data through to the epistemic boundary of what evidence alone can establish. Original diagram prepared for Budding Forensic Expert.
Mandatory Evidence Status Table
| Information Investigators Want | Can Blood Potentially Provide It? | Method/Approach | Current Forensic Status | Major Limitation |
|---|---|---|---|---|
| Presence of blood | Yes | Presumptive colorimetric/chemiluminescent tests | Established | Moderate specificity; false positives documented |
| Human origin | Yes | Immunoassay (ELISA/precipitin); species-specific DNA markers | Established | Cross-reactivity risk with closely related primates |
| Biological source (fluid type) | Partially | mRNA/miRNA typing; proteomic peptide panels | Validated in specialised contexts | RNA degrades quickly; proteomic panels not universally standardised |
| DNA profile | Yes (quantity/quality dependent) | STR profiling; probabilistic genotyping | Established, with caveats for mixtures | Low-template and mixture interpretation carry statistical assumptions |
| Contributor information (sex, ancestry, appearance) | Partially | Forensic DNA phenotyping (SNP panels) | Validated for some pigmentation traits; emerging for others | Population dependence; weaker for intermediate/admixed traits |
| Chronological/biological age | Partially | DNA methylation clocks; miRNA models | Emerging research | Margin of error of several years; environmental confounds |
| Chemical/drug exposure | Yes, with major caveats | LC-MS/MS on dried blood spots or liquid blood | Established for detection; interpretation limited | Post-mortem redistribution distorts concentration-timing inference |
| Bloodstain age (time since deposition) | Research potential; not operationally reliable | UV-Vis/ATR-FTIR/Raman spectroscopy; RNA/protein degradation models | Emerging / not court-ready as a single method | Highly environment- and substrate-dependent |
| Physiological/disease state | Largely no, for unidentified samples | Clinical biomarker analogy; proteomics | Exploratory research | No forensic diagnostic baseline for source-unknown samples |
| Event reconstruction | Only via integration with other evidence | Bloodstain pattern analysis; case-level Bayesian reasoning | Contested reliability | Documented inter-analyst disagreement and error rates |
| Identity (named individual) | Only via comparison to a reference | STR profile comparison/database search | Established | Requires a reference sample or database hit; statistical, not absolute |
| Timing of an event | No, from the sample alone | N/A — requires scene/witness/digital context | Beyond biological evidence | Deposition time ≠ collection time ≠ event time |
| Intention or motive | No | N/A | Beyond forensic biology | A legal/psychological question, not a biological one |
Direct vs Inferred Information Table
| Question | Directly Measurable From the Sample? | Requires Interpretation? | Requires Additional Evidence? |
|---|---|---|---|
| Is this substance blood? | Largely yes | Minimal | No |
| Is it human blood? | Largely yes | Minimal | Rarely |
| Whose DNA profile is present? | Yes (profile); comparison requires a reference | Moderate (especially mixtures) | Yes — reference sample or database |
| What is the contributor's likely appearance/ancestry? | Partially | Substantial | Population reference data |
| How old is the stain? | Partially, with wide uncertainty | Substantial | Environmental/substrate context |
| How did the blood get here? | No | Substantial | Yes — transfer/persistence reasoning, scene context |
| What activity caused the bleeding? | No | Substantial | Yes — scene, witness, pattern evidence |
| Was the contributor present during the crime? | No | Substantial | Yes — full case-level evidence integration |
| Did the contributor commit the offence? | No | Not a forensic-science question alone | Yes — the entire legal process |
Major Scientific Limitations (Summary)
- Presumptive test specificity remains a persistent source of false-positive risk that confirmatory testing only partially resolves.
- Probabilistic genotyping software, while a genuine advance over binary allele-counting, embeds statistical assumptions that experts disagree on how transparently courts currently receive them.
- No single bloodstain age-estimation technique has yet achieved environmentally robust, cross-substrate operational reliability.
- Bloodstain pattern analysis rests on sound physics but documented inter-analyst variability and susceptibility to contextual bias in practice.
- Forensic DNA phenotyping accuracy is markedly uneven across trait categories and population groups, with admixed and non-European-reference populations comparatively under-studied.
- Post-mortem redistribution complicates the interpretation of toxicological findings in deceased-donor blood far more than in living-donor samples.
- Cognitive bias from task-irrelevant contextual information is a documented, measurable influence across multiple forensic biology and pattern-interpretation tasks.
- India-specific population and environmental validation data are limited across nearly every emerging technique discussed in this article.
Future Research Directions
- Cross-laboratory, cross-substrate validation studies for bloodstain age estimation under genuinely uncontrolled environmental conditions.
- Expanded population reference panels for forensic DNA phenotyping that include South Asian and other under-represented population groups.
- Continued black-box reliability testing for bloodstain pattern analysis, alongside implementation of bias-mitigation protocols such as Linear Sequential Unmasking.
- Multi-marker (RNA + methylation + protein) integrated assays that may improve on any single marker type for age and body-fluid estimation.
- Transparent, standardised reporting frameworks for probabilistic genotyping software outputs, addressing the concerns raised in recent comparative validation literature.
- India-specific validation studies across the full range of techniques discussed in this article, particularly given the country's recent statutory expansion of mandatory forensic crime-scene involvement.
Conclusion: Where the Information Ends and Inference Begins
A crime-scene blood sample can, across its many molecular and physical layers, carry a genuinely extraordinary quantity of forensic information — biological, genetic, chemical, and physical. Whether that information can be responsibly turned into an investigative or evidentiary conclusion depends on sample quality, the specific analytical method used, how thoroughly that method has been validated, the sample's environmental history, the skill and transparency of interpretation, the surrounding case context, and a disciplined consideration of alternative explanations at every step. None of these dependencies are hidden or unusual within forensic science; they are, in fact, the ordinary conditions under which all rigorous science operates. What distinguishes forensic biology is simply that its conclusions carry direct legal consequence, which makes the discipline's obligation to state its limitations honestly considerably higher than in most other scientific domains.
The right question to close on, then, is not simply how much information a bloodstain contains. It is how much of that information forensic science can currently recover with demonstrated reliability — and how far an investigator, an expert witness, or a court can legitimately extend that recovered information into a claim about what actually happened.
