In his famous 2005 paper, the epidemiologist Christopher Wild was one of the first to give the complementary concept to the genome a name: the exposome. According to Wild, the exposome includes “life-course environmental exposures (including lifestyle factors) from the prenatal period onwards”.³⁷ Gary Miller and Dean Jones extended the concept of the exposome in 2014, shifting the focus of the exposome beyond solely the exposures.³⁸ They expanded the concept adding the aspects of diet, behavior and endogenous processes, with a particular focus on the biological responses to exposures. Even genetic and genomic alterations serve as evidence of past exposures.³⁹ The resulting shift in exposomics research from exposure-focused (introduced by Wild)³⁷ to more metabolomics focused approaches³⁸ to capture biological endpoints has been accompanied by several changes. Of course, the study of internal metabolic changes and the search for biomarkers in exposomics research is highly necessary, but not the sole goal of exposomics. The last years showed a trend to using increasingly expensive resources and sophisticated techniques, rather than focusing on the initial research question. Research is driven by metabolomics approaches using databases that contain mostly “known” knowledge, necessary to ensure workflow efficiency, yet at the same time trying to find new discoveries and improve the understanding of metabolic pathways. However, with the rapid pace of innovations in the age of the Anthropocene, new synthetic compounds (and their related transformation products) are appearing at an alarming rate with up to 20 million new registrations a year,⁴⁰ which can cause new problems and influence the human metabolism very differently. Although it is possible to create dynamic workflows and databases to profit from “new” knowledge generated via high throughput exposomics (eg, PubChemLite for Exposomics⁴¹), this requires concerted efforts at FAIR and Open data exchange,^42,43 for which resources and infrastructure are still under development. The exchange of information will be vital to support reinterpretation of older data (eg, retrospective screening^44,45) or large community initiatives for large scale discovery such as the Global Natural Products Social Molecular Networking (GNPS) ecosystem.⁴⁶ However, while modern analytics and monitoring techniques can help capture relatively recent perturbations in the exposome, even using natural archives to an extent to investigate past pollution, there is a limit to the availability of suitable samples. The historical exposome concept can give additional perspectives to this technologically driven bio-exposome. Looking at environmental history as such, especially focusing on developments in the Anthropocene, can show many interconnections of environment and health, where expensive technological studies using biological samples of today may fail to find these connections. Historical archives can, for instance, reveal if currently perceived extreme events such as drought or flooding (relative to written records of ∼100 years) were in fact observed centuries earlier. Closer to the exposome, anecdotal, written documents (letters, newspaper articles) can also reveal the catastrophic side-effects of industrial pollution in factories, which is only partially captured in scientific literature (and if covered, often in foreign languages).⁴⁷

One way to approach the exposome from a more holistic perspective is to embed it into the wider framework of what has been called “the environment” since the early post-war years. As demonstrated by the three environmental historians Paul Warde, Libby Robbin, and Sverker Sörlin in their book “The Environment. A History of an Idea”,³⁰ the term gained prominence as a political concept as a result of the success of popular science writers such as William Vogt (“Road to Survival”, 1948)⁴⁸ and Rachel Carlson’s best-seller “Silent Spring” (1962).²⁹ Although the term had been around for more than a century (“milieu” in French, “Umwelt” in German), a transformation of meaning from “a world where man was molded by environment to him being able to alter the nature of his world”^{30 (p. 8}) only occurred with the “environmental revolution” in the 1970s.⁴⁹ Sparked by new academic interventions such as the “Limits to Growth” report to the Club of Rome in 1972⁵⁰ and the first “oil crisis” in 1973, ecological thinking became a mainstream concern for both scientists, environmental or ecological movements, and world politics. The problem of “the environment” became both scaled (ie, local problems of pollution or waste were interpreted as a subset of a planetary issue) and the result of a complex interplay of past, present, and future temporalities.²¹

To handle the growing complexity of issues at stake, new concepts such as the “ecosystem” were put forward to highlight the holistic nature of environmental changes. Inspired by the post-war boom of “systems theory” and “cybernetics”, and long before climate science would emerge as a new interdisciplinary field, even Eugene and Howard Odum introduced the metaphor of balance between the living (organic) and nonliving (abiotic) environment in their textbook “Fundamentals of Ecology” (1953).⁵¹ Ecosystem science became an important framework for setting up large-scale international research projects, shaping the thinking of the environment as a dynamic interplay between “nature” and “nurture” in a system called “earth” at a planetary scale.^{30 (pp. 154–158)} Since the introduction of the concept of the “Anthropocene” by Paul Crutzen and Eugene Stoermer in 2000,⁸ the human factor has entered the equation: according to recent scholarship, both the living and the nonliving environment has been radically affected by human activity on earth, crossing planetary boundaries.^52-54 With the human being acknowledged as an ecological factor in history (as had been suggested by historian long before),⁵⁵ the concept of environment as a tripartite dynamic system has emerged, combining the “natural” environment with the “human built” or socio-technical environment and the subjective environment of the individual.

This “holy trinity” of natural, individual, and social factors that form “the environment” suggests that the physical surrounding in which life takes place, which is comprised of the atmosphere, the biosphere, the hydrosphere and the lithosphere,^{56 (p. 14}) is fundamentally interconnected with human activities on earth. We cannot grasp the full complexity of one element without considering the impact of the others. The interconnected, dynamic relationships render “the environment” an incredibly complex system that seems hard to grasp—even from a truly interdisciplinary perspective. Yet it is this “maelstrom” of temporalities, scales, and uncertain causalities that challenges mono-disciplinary approaches and interdict mono-causal scientific explanations.⁵⁵ What is needed are new questions, concepts, and narratives aiming at forging an understanding of past and present environmental challenges, starting with a new search for evidence in and significance of existing data from both natural and social archives. Combining evidence from historical and natural sciences and thereby bridging the “two scientific cultures” is a key condition for performing historical exposomics research.

The search for significance is a common factor across all disciplines. Experts from the fields of humanities, natural and social sciences aim to make significant contributions in their areas of research. In a more general sense, significant contributions have the quality of being important, of being worthy of attention because they answer relevant questions and have meaningful results. In practice, the mindset, practices, and evaluation methods of assessing significance are often specific to each discipline, with a diverse range of views also present within each discipline.

One reason may be the nature of the information sources in each discipline. Focusing on the fields of history and natural sciences, a historian works more frequently with primary sources such as historical manuscripts or photographs, while a researcher in natural sciences often works with numerical data resulting from measurements. The inherent difference in the nature of the sources, as well as the a priori impossibility of making measurements in the past or repeating an experiment, lead to different approaches to the assessment of significance. Additionally, an apparent widespread dualism can be perceived, where natural sciences more often try to prove a theory that can be generalized for a larger population, whereas history is more frequently concerned with specific knowledge about a particular fact or occasion, with the consequent differences in the researcher’s approach to significance.

Statistical hypothesis testing is an inference method widely used in natural and social sciences to determine a possible conclusion between two hypotheses. A null hypothesis is defined against an alternative hypothesis and the P-value is defined as the probability, calculated under the null hypothesis, that a test statistic is as extreme or more than its observed value. If the P-value is less than the selected significance level, then the null hypothesis is rejected. Traditionally, a P-value < 0.05 has been considered a reasonable significance level, but in some fields such as genomic studies, more stringent levels are adopted. The criticism of this method is extensive across fields of research,^57-60 including technical limitations such as the difference between the characteristics of the scientific data as opposed to the assumptions upon which the significance tests are defined, sampling issues regarding size and randomness, the arbitrary level of significance, the dichotomous reject/not-reject, the misinterpretation of P-values and the lack of reproducibility. However not only technical issues have been raised, but the actual scientific value of the tests has repeatedly been questioned, pointing out weaknesses in the use of isolated tests, the difficulty to generalize results and to integrate previous knowledge, causing erroneous scientific reasoning.

Numerous options have been proposed over decades of research, from the use of stricter P-values to the total abandonment of statistical significance, as well as a wide range of alternatives,^61-63 but the statistical significance in hypothesis testing is still a widely used method due to its intuitive interpretation, ease of calculation with existing tools, and facility given to the choice of the research path with yes/no questions. Bayesian approaches, older than frequentist statistics, gained popularity with the advances in computational methods. Bayesian methods depend on a prior and on the probability of the observed data, allowing the sensitivity of the experiment result to be measured for different priors.^64,65 In Bayesian inference, the uncertainty or “degree of belief” with respect to the parameters is quantified by probability distributions.

Controversy about statistics at a more general level includes the opposition of two attitudes to the question of reality, one realistic or objectivist (ie, pre-existing); and the other, relativistic or historicist (ie, constructed),⁶⁶ very present when bringing together history and natural sciences. In Trust in Numbers, Theodore M. Porter challenges the ubiquity of quantification in the sciences of nature, highlighting that only a small proportion of the numbers and quantitative expressions “make any pretense of embodying laws of nature, or event providing complete and accurate descriptions of the external world”.⁶⁷

There is an increasing interest in qualitative research methods across disciplines, such as research techniques that rely on non-statistical or numerical methods of data collection, analysis and evidence production. Qualitative research allows to retain complexity and nuance through data collection methods that adapt to the context and make it possible to explore emergent issues; and they present a reflexive approach that acknowledges the perspective of the researcher in the process.⁶⁸ They are used in the social sciences and the humanities, but they can also complement quantitative approaches in the natural sciences.⁶⁹ Qualitative research can be used independently to uncover topics that are not amenable to quantitative research. Furthermore, it can be used as the preliminary of quantitative research, that is to uncover ambiguities and misunderstandings regarding terms and definitions, to help understanding the reasons behind certain results, or to validate quantitative research by providing a different perspective on the same phenomena, sometimes forcing major reinterpretations of quantitative data.⁷⁰ Despite the many potential applications, qualitative methods are often evaluated according to quantitative measures of rigor and dismissed as unscientific and unreliable.⁷¹

Criteria of significance in history includes two main questions: what is important to learn about the past and how do historians know what they know. The first refers to deciding which events or people resulted in a change that had deep consequences. The second is about the historical interpretation based on inference made from sources: what questions turn a source into evidence, who created it, when and for what purpose, what is the historical context and w the inferences can be corroborated.⁷² Partington⁷³ identifies three criteria of significance. First, the importance to people in the past “if we use the egalitarian principle of counting heads to establish priorities here we are in difficulty, because we have unequal access to the opinions and judgements of different social and ethnic groups […].”⁷³ Second, objective criteria that includes profundity, as whether an event profoundly changed people’s lives; quantity, as the number of lives impacted; and durability of the event in time. Thirdly, criterion is relevance, or how an event contributes to an increased understanding of the present.

An interdisciplinary approach to historical exposomics blurs the boundaries between discipline-specific approaches to significance and makes it possible to study the past from new perspectives. Historical sources like pollution measurements included in official reports, complaint letters from the citizens about the dust from the factories and contaminated water, newspapers publishing about new labor laws, innovative industrial techniques, or disease outbreaks, can be used to test hypotheses about pollution concentration or the impact on health in the past, while assessing at the same time for historical significance. The historical archives are filled with scientific and non-scientific, numerical, and non-numerical sources, that together with current scientific knowledge and measurements about water systems or use of chemicals, allow to simulate and validate hypothesis about the past. The dualism “specific” versus “universal” between science and history is replaced by a combination of close and distant reading from a historical and scientific point of view adapted to the source and specific analysis, which allows access to much more information, to analyze it from different points of view, and to use multiple approaches to assess significance.

A historical exposomics approach requires the combination of a great variety of historical and current data. As shown here using data collected for the research project “LuxTIME” as an example, the information gathered comes from both “natural” and “social” archives. Natural archives are physical objects which have been collected, processed and deposited in the environment under natural circumstances (ie, without the interference of the human species), and preserve information about the characteristics of the surroundings from the time and place where deposition took place.⁷⁴ Through various analysis techniques, this stored information can be extracted from the archive, enabling the reconstruction of environmental conditions far beyond the period where direct measurements have taken place. This helps determine the baseline conditions at the time, as well as spatial and temporal changes, allowing researchers to relate these to natural or anthropogenic influences.⁷⁵

Both natural sciences and humanities use data in their research, whether they are historical or modern texts, measurements from observations or experiments, elements of a photograph, time series, geographical data or many more. Data is the connecting element in interdisciplinary research. Various disciplines often share data analysis methods, but also share the same challenges, such as how to discover what is relevant for their specific research question when dealing with a large volume of data, how to validate the data, or how to communicate the results to readers according to their specific needs. They also face fundamental differences in the definition and use of data and the required communication objectives. Data visualization supports multiple scenarios including discovery and communication. It helps to navigate large volumes of data, whether to perform an initial exploratory data analysis to define the line of research, or to make discoveries that form part of the project results. It is also essential for communicating results, which allows creation of the desired user experience in each case.

Especially in the context of interdisciplinary projects, it is important to understand the different approaches to data visualization since a combination of these might be required for the analysis of data from the point of view of different disciplines. In the case of the natural sciences, data visualization is generally used as a tool to simplify a complex topic, to make data-driven decisions about the best path of research, to build upon previous scientific knowledge by integrating references and benchmarks, and to communicate a clear message to an audience. Scientific data visualization is largely based on statistical graphics which are characterized by the principles of abstraction, reduction, standardization, representation, and legibility. The user is expected to be familiar with the visual elements—the statistical charts, for example, bar charts, line charts—and to ask predefined questions to get formatted answers, to apply filters, and to drill down to predefined levels of aggregation. In the case of the humanities, the use of data visualization often aims at showing the complexity of the subject of research, to engage the user in the interpretation and to present multiple narratives. The data visualization is defined by its granularity, specificity, and full coverage.

To study historical exposomics, data visualization techniques such as concept maps can be used to understand the specific contributions of each field and the joint area of work, see Figure 4.⁸⁵

During the exploration of the data, the use of data visualization allows identification of the types of data available—time series, geographical, texts, images—the geographical area and the period covered, the variety of sources—historical archives, sampling campaigns, web archives—the domains—air, water and soil pollution, demographics, health, industry, urbanism, green and blue areas—and the percentage of sources digitized, among others. This facilitates the identification of gaps, areas of interest and future challenges. Lastly, one of the most frequent use of data visualization is to share the research output. A combined approach to data visualization, scientific and humanistic, allows integration of representative analyses to understand the general impact of the main environmental exposures on health and exploration of the different narratives by integrating individual cases, when relevant.⁸⁵

In studying historical evidence of human traces on the environment by applying both scientific and hermeneutic methods of interpretation, LuxTIME aims at contributing to a new form of data-driven scholarship on the Anthropocene by experimenting with digital forms of historical storytelling, such as animations of dust pollution in Esch-sur-Alzette over nearly one century (1911–1996) shown in the website “minett-stories” and in Figure 5.

The animation that was produced within the virtual exhibition “Minett Stories”¹³²—an interdisciplinary public history project in the framework of the “Esch22/Capital of Europe” initiative,¹³³ was the result of a collaboration between historians, metallurgy experts and chemists. Based on a data set including information about the production output of the Belval steel plant throughout the 20th century, the type of production process (eg, LD-AC or Thomas for steelmaking), the typical usage of filtering systems, and the prevailing wind directions in Luxembourg over time, the animation offers a plausible approximation of how the dust emissions of the Belval iron and steel complex might have affected local air quality over the decades. Even if the animated map can only offer a hypothetical approximation rather than an exact reproduction of past reality, such temporal visualizations can trigger our historical imagination and open new paths for further historical investigation.