5. Probing sources and methodologies
SIGNPOST: Relevant to all researchers
A salute to historical detectives everywhere.
5.1 How good are the sources?
Historians are very often asked: ‘What are the sources?’. Yet that query should really be paired with: ‘Are the sources good enough for your project?’. Using evidence from the past entails understanding its strengths as well as its biases, limitations and omissions. There’s thus a constant mental dialogue between analysing sources and building conclusions. Good historians, like good detectives, need to start somewhere but also keep an open mind. So they should both follow and challenge the evidence.1 If, during a multi-year project, researchers have not either confirmed a previously shaky view or formulated a valid new one, then they have not been doing their job. Historians need constantly to revisit their assumptions and, if need be, to adapt their views as things develop. Such refinements are the essence of historical scholarship.
Interrogating the sources does not imply that evidence from the past is being blamed or belittled. That’s not the intention at all. Historical sources simply are as they survive – or as, in some cases, they can be reconstructed. Yet they all need to be used with real care. The golden rule is to play fair with the evidence.
As already stressed – and it is worth stressing again – falsifying data, misquoting sources, selecting only positive information, hiding unfavourable evidence or mangling research techniques are all supreme academic sins.
Critical reflections upon the nature of the evidence usually appear near the start of all research studies. Sometimes a source audit may be summarized briefly, while at other times, it may command an entire chapter or more. Such commentaries are not pedantry. In fact, exploring the strong and weak points of specific sorts of sources can turn into a fascinating exercise in itself. And, once complete, the audit establishes the basis for the study that follows. Indeed, most supervisors ask researchers to start with this process precisely to produce an overview that sets the scene both knowledgeably and critically.
At this point, it’s worth emphasizing that researchers should think laterally and imaginatively about how to deploy and interrogate the evidence. Materials need not only be used for the purposes for which they were created. For example, historians of religious change in sixteenth-century England examined the preambles to ordinary citizens’ wills. The aim was to detect a shift in expressions of popular adherence from Catholicism to Protestantism. Of course, some testators might have relied upon friends or clerical advisors when making wills; but, in such cases, it was the helpers who were reflecting the shift. On the other hand, will-making may have remained throughout a matter of unchanging cultural ritual.2 Hence the evidence was suggestive but not conclusive. It’s a classic case of rival interpretations of a change for which there was no one single definitive source. But it’s also a creative way of considering how big policy changes at a national level may have impacted upon daily lives.
5.2 The preliminary source check
Provenance
It’s always worth starting with a basic check. What is the provenance of any given category of source material? Who created it and why? How has it survived from its original state through to the present day? How well authenticated is it? Has it been amended or changed over time? (Usually yes.) Has the entire collection any in-built biases – perhaps reflecting past racist and/or sexist attitudes?3 Was the evidence amassed for a particular purpose – often as an archive of record, but sometimes as a collection for propaganda purposes? All these resources provide real and effective documentation, but they are not neutral lenses upon the past.
Usually all items that have been assembled in public archives, libraries, galleries and museums have been authenticated before they are incorporated into a collection. That process does not mean that their interpretation is beyond debate (far from it), but it does confirm that all materials are genuinely what they purport to be. However, a small proportion of mis-attributed items or subsequent fabrications may also creep into the record. There are also processes of restoration, whose impact needs to be recognized. On a grand scale, visitors to present-day Bruges4 or the palace of Knossos in Crete5 might think that they are viewing a well-preserved medieval city or a Cretan archaeological marvel, but they are actually witnessing the impact of zealous (some might say over-zealous) restoration by nineteenth- and twentieth-century enthusiasts. Their handiwork is thus, at one and the same time, real historical evidence and ‘fake’.
If there is any doubt, the matter should be checked and rechecked. The entire journey of all materials from formation to preservation to research usage needs interrogation. The same rule applies to material in private holdings. It is unusual for individuals to go to immense trouble to fabricate authentic-seeming sources. But it is easy enough for family folklore to attribute spurious age and authority to the disorderly contents of an attic.
Cases of documentary and archaeological fakes are all too well known. These are sometimes perpetrated for motives of financial gain – or to substantiate a controversial theory. In 2005–6, that allegedly prompted the historian, Martin Allen, to insert twenty-nine fabricated documents, relating to Britain’s policies in the Second World War, into The National Archives.6 Researchers should accordingly beware! The first requirement is to check (however swiftly) the provenance of all sources.
Furthermore, the frequent repetition of items of knowledge as known ‘facts’ does not automatically guarantee their accuracy. At times, errors can slip through the net. Falsehoods can be repeated so frequently that they are assumed to be true, until exposed as erroneous, thereby seriously wrong-footing all the experts who had relied upon their veracity. An example can be found in the alleged popular riots against calendar reform in England in September 1752. Many historians had repeated the story. But Robert Poole’s meticulous research revealed that the popular riots, when people allegedly shouted ‘Give us back our eleven days’, did not happen.7 They were a later myth, built on assumptions that the poor were superstitious, credulous and automatically hostile to change. Such phoney facts are known as ‘factoids’, in an eloquent term first coined by Norman Mailer in 1973.8 So again, the message is simple. Facts too may prove dubious; and, if in any doubt, researchers should check and double-check – again.
Reliability
A source or group of sources may still be authentic but not necessarily reliable in the sense of being precise or accurate. Materials from the past have no historic duty to be anything other than what they were. A song about ‘happy times’ is no proof that times were actually happy, either when the song was first written, or in all the subsequent eras when it was sung again. Yet it’s significant that songs to happy times are written – and repeated in the repertoire. Their contents, their music and their performance history (if known) can all be decoded. In that way, the history of songs constitutes a very fruitful field of enquiry, provided that it is recalled that survival rates for this sort of ephemera are patchy – and that song lyrics are not taken as the equivalent of declarations on oath.9 That latter point applies to much subjective documentation from the past. Statements which are not made on oath have no obligation to be accurate; and even sworn affidavits may lie.
Generally, it’s enough for historians to flag the difficulties in interpreting subjective testimonies, and to take such challenges into account when interpreting the material. Such assessments are the stuff of interpretation. The subjective nature of many documentary sources is not an insurmountable problem as such, but it is an issue which needs to be handled with real thought and care. How to deal with often complexly flawed evidence is one of the primary challenges for historians.
Another problem may also emerge when ‘primary’ sources are subsequently edited – and when the originals have disappeared, thus removing the chance for checking. For the most part historians have to take editorial accuracy on trust. Yet, if there is any doubt, then they should immediately accept that the source is no longer a valid ‘primary’ source but a doctored, later concoction. One classic example was the removal of religious fervour from the mid seventeenth-century Memoirs of Edmund Ludlow by the late seventeenth-century editor, who was a militant rationalist. Only very much later was a manuscript version discovered. A cross-check between the two versions was salutary. It showed that the editor had given the Memoirs a markedly secular tone, which was long, but wrongly, accepted as authentic.10
Digital sources have also introduced further problems. Take the example of the Burney Collection of Seventeenth- and Eighteenth-Century Newspapers. The process of making them available online has built in sundry biases. First amassed in the late eighteenth and early nineteenth centuries, the collection was acquired by the British Library in 1818, and slowly augmented by additional sources. But from the start the collection was substantially biased towards London. For preservation reasons, it was then microfilmed in the 1950s, greatly enhancing its availability. However, those enthusiastically searching the contents made only little acknowledgement of their metropolitan bias.
Then, in the early twenty-first century, the collection was translated into a new digital version via a process called Optical Character Recognition (OCR). This automatically generates a searchable electronic text.11 But, sad to say, the OCR of print from the seventeenth and eighteenth centuries is notoriously prone to error. Less than 50% of semantically significant words in the online version of the Burney Collection are transcribed accurately.12 And that figure rises for text printed in italics or in tabular form. The Burney Collection is so substantial that most keyword searches will still produce useable results. But historians must get to grips with the problems generated by its metropolitan bias and with the current technological limitations upon word recognition. Consequently, the universal motto is: researchers be aware!
Typicality
With all sources, it’s also helpful to pose the question as to whether their evidence is likely to be commonplace or highly unusual. Again, it doesn’t matter which it is, as long as the implications are fully taken into account. Otherwise, there is a danger of generalizing from something that is in fact a rarity; or, conversely, of taking something as exceptional which was actually commonplace.
Assessing typicality is not always easy, especially in the case of obscure and fragmentary materials. Yet it’s always helpful, wherever possible, to check one source against many comparable examples, to gain a sense of the genre. And it’s equally good to acknowledge difficulties when using the said evidence.
Keeping a sense of proportion is particularly relevant when supporting references for something previously unknown can be found relatively rapidly by keyword searching of digitized data. It’s good that such methods can both find and contextualize a new piece of information. For example, the unorthodox historian E. P. Thompson – a great searcher after new sources – startled his fellow researchers in the 1970s with a lecture on eighteenth-century wife sales. These de facto popular divorces, forbidden by both church and state, were very rarely mentioned. Yet they did occur. Once Thompson had alerted the research community, then others found that their eyes were opened to new cases, many being recorded in the provincial newspapers. The topic was thus well and truly brought in from the cold.13 Yet it is still helpful for researchers to indicate that wife sales were far from commonplace, chiefly because of their problematic legal status. The couples settling their matrimonial affairs in this manner (usually by joint agreement) were taking a calculated risk to gain public acknowledgement of their changed status. A sense of scale, in other words, reveals the full meaning.
5.3 The close source critique
Context
Having looked generically at a source or group of research materials, it’s then useful to make precise checks. Step one advises that all evidence be put into its context of time and place. One example makes the point. Finding a sheet of paper inscribed with the words ‘William, son of John Shakespeare’ would not get a researcher very far. But locating them in the parish book of Holy Trinity, Stratford-upon-Avon, dated 26 April 1564, provides good evidence of the baptism of the world’s most famous William Shakespeare.
The document in question contains four words in Latin: ‘Gulielmus filius Johannis Shakspere’. A later hand marked the entry with three inked crosses. It was an endearing sign of research excitement, although today it would rightly attract the wrath of archivists. Hence it can be reliably assumed that Shakespeare was born some time very shortly before his baptism on 26 April. The precise date, however, remains unknown. (A patriotic tradition dating back to the eighteenth century ascribes it to 23 April – St George’s Day.)
Style or register
Step two requires a full understanding of the source’s characteristic style or ‘register’, in the terminology of literary scholarship. At a very basic level, there are obvious differences in written texts between fiction and non-fiction. Poems, stories and songs are not intended to be taken literally. And within the ranks of non-fiction, there are many different types of writings, and levels of specificity. Private thoughts expressed casually, in (say) letters and diaries, do not necessarily constitute people’s final considered views.
Researchers thus have to allow for human variability. Guidebooks also helpfully review the common characteristics of personal sources, such as autobiographies,14 diaries,15 reported speech16 and letters.17
Again, the historic documentation relating to William Shakespeare provides a salient example. When his will bequeathed to his wife Anne Hathaway their ‘second best bed’, he was not comparing her to a summer’s day. He was leaving her a specific item of household furniture. It can be debated whether the legacy was a considered snub or a tender personal testimonial or a utilitarian disposal of family assets or just a casual after-thought (given that this bequest was visibly a late addition to the will, interpolated between two lines of an already written text).18 But the testator’s motivation was not specified. In other words, the legal register provided terse wording for a specific purpose, differing utterly from the poetic register, which can be anything from closely observed to rapturous to nonsensical.
It’s thus for the researcher to determine how a given source can or cannot be used (and to defend that decision, if challenged) – and that point applies whatever the nature of the surviving evidence from the past.
Contents
After all these necessary preliminaries, researchers can at last savour the full contents of the sources in question. Every last detail may be important. As PJC’s former research supervisor Jack Fisher used to say: squeeze every last drop of juice from the lemon.
Possibilities, however, always remain bounded by the availability of evidence – and by the time and methods available for analysis. Again, it’s worth repeating those salient points. Some things are genuinely lost from history. So, without fresh finds, Shakespeare’s actual birth date will remain unknown; and his attitude to his wife, when making his will, also remains a matter of conjecture.
Of course, it’s always possible that new discoveries may be made. A good place to look would be within the only partially catalogued sixteenth-century legal records. Meanwhile, all new finds and claims about Shakespeare will be subject to extra-stringent checking and criticism from the world’s array of enthusiastic bardologists.19
Considering the question of missing sources more broadly, some big fields of research will entail finding careful mechanisms to work around major absences. When the British government gave up many of its colonial possessions in the 1950s, it destroyed many historical records – literally burning them or dumping them in the sea.20 The motivation may have been either to avoid the effort and cost of preservation; or, more deviously, to sidestep the danger of any subsequent allegations of political malfeasance.
Either way, historians analysing the colonial era will need to proceed without those documents, but in the knowledge that they once existed. As in so many cases, ingenuity is the best response to challenge.
5.4 Choosing appropriate methodologies: general principles
‘Methodology’ is an abstraction which refers to the different research methods, whether straightforward or technically complex, which all historians adopt.21 It’s high time to demystify the term, which is not especially lovely but is useful. Earlier generations of British historians tended to bristle at the very word. They associated it either with formal sociology or with Teutonic abstraction – or both.
Nowadays, however, both the concept and the word have worked their passage into scholarly usage. It’s true that many historians still stick with the more accessible phrase ‘research methods’. Yet references to ‘methodology’ are no longer treated with incomprehension or denial. Instead, just as it’s generally agreed that researchers should play fair with their sources, so there’s a consensus that they must play fair with their methodologies too. Put simply, how have the data been explored/tested/processed? And are the procedures appropriate? The rise of new digital methods (discussed in detail in chapter 10) is also encouraging researchers to re-examine more traditional approaches and to develop new ones.
Much the commonest approach is for researchers to read/view/listen to many different sources; to evaluate them; to compare and contrast them together; and to use them to construct conclusions, in the context of the secondary literature. That common-sense process is sometimes derided as theoretically ‘naive’. Yet it has the great advantage of being robust, explainable and readily communicated to a wider audience. Indeed, history researchers generally do not apply ready-made theories (especially as there are often rival variants); and, if they do follow a particular ‘line’, then they still have to test it critically too. The crucial feature, in every case, is to explain clearly what approach has been taken and why.
There are also a myriad of technical procedures, designed to enhance the analysis. To take just one example, the process of radiocarbon dating allows researchers to establish the production date of material objects which do not speak for themselves.22 All well and excellent. But again, readers should be invited to share a rigorous methodology critique, along with the source critique.
In shorthand, the research methods are sometimes divided into a polarity, known respectively as ‘soft’, meaning qualitative and subjective, and ‘hard’, meaning quantitative and impersonal.23 But those terms are unfortunately loaded. ‘Soft’ techniques may seem too easy and too cosy. ‘Hard’ approaches, by contrast, rely upon mathematical/quantitative techniques, also known as ‘cliometrics’.24 Typically, they pose precise questions and seek quantified answers. The side-implication is that such exercises enshrine not just ‘rigour’ and ‘accuracy’, but also ‘difficulty’. And there is some truth in that last characterization, in that advanced cliometric calculations generally require a degree of technical sophistication which is beyond most historians.25
Nonetheless, precise statistics, where available and appropriate, are absolutely invaluable. In economic and demographic history, they usefully reveal both scale and trends – and they can be used to calculate the relationship between different variables, allowing always for the null hypothesis that no meaningful relationship exists.26 Another new process that is gaining popularity within digital humanities is known as ‘distant reading’, which entails counting words and their associated meanings. (For more, see chapter 10.)
Yet the conventional ‘hard/soft’ dichotomy is ultimately misleading in that all historical researchers seek rigour and accuracy. As a result, many practitioners, like most social scientists today, use a mixture of qualitative and quantitative approaches, chosen as relevant to the subject in hand.27
What matters, therefore, is the aptness of each approach, whether qualitative, quantitative or mixed. All methods should pass three tests: first, that the sources are good enough to carry the analytical superstructure; second, that the techniques are not time-consuming beyond the remit of the project; and third, that there is no risk of expending time and effort to reveal an answer that was obvious from the start.
Specifically, on that point, it’s vital to ensure that no conclusions are inadvertently built into the methodology or the system of data classification. Unchecked assumptions about gender, race and/or class, for example, can produce distortions by inadvertently inserting conventional prejudices into the organizational categories.28 Instead, methodologies must enhance, not pre-determine, the analysis.
5.5 Summary: the open-ended quest
Studying history is genuinely fun and incomparably stimulating – and that’s what keeps researchers going. Finding, not finding, scrutinizing, assessing, comparing, collating and interpreting evidence from the past can entail hard mental graft, which never seems to end. (The past is infinite.)
Not every project will yield the purest gold. But there is always scope for great discoveries and new interpretations. Or, equally, breakthroughs can be built upon many small steps. In all cases, researchers have to explain themselves to others – and, better still, to convince them. The process should always be dialogic; the quest open-ended.
Seeking fresh knowledge means entering into the unknown. As a quizzical dictum, often attributed to the great exploratory physicist Albert Einstein, mused: ‘If we knew what it was we were doing, it would not be called “Research”, would it?’.29
1 On that two-way theme, see Anon., The Historian as Detective: Essays on Evidence (London, 1969); plus R. B. Browne and L. A. Kreiser (ed.), The Detective as Historian: History and Art in Historical Crime Fiction, Vol. 1 (Bowling Green, Ohio, 2000); Browne and Kreiser (ed.), Detective as Historian, Vol. II (Newcastle upon Tyne, 2009).
2 See C. J. Litzenberger, The English Reformation and the Laity: Gloucestershire, 1540–80 (Cambridge, 1997), pp. 168–78; E. Duffy, The Stripping of the Altars: Traditional Religion in England, c.1400–c.1580 (London, 1992, 2005), pp. 515–21; J. D. Alsop, ‘Religious preambles in early modern English wills as formulae’, Journal of Ecclesiastical History, xl (1989), 19–27.
3 T. Shellam and J. Cruickshank, ‘Critical archives: an introduction’, Journal of Colonialism and Colonial History, xx (2019), Project MUSE, doi: 10.1353/cch.2019.0017.
4 See J. McDonald, Bruges: Historic Walking Guides, ed. Z. Wildsmith (Durham, 2009); ‘Just How Medieval Is Bruges?’ (2020): <https://en.tripadvisor.com.hk/ShowTopic-g188671-i448-k12710656-o10-Some_travelers_claim_that_Bruges_is_a_fake_medieval_city-Bruges_West_Flanders_Province.html> [accessed 29 April 2021].
5 J. A. MacGillivray, Minotaur: Sir Arthur Evans and the Archaeology of the Minoan Myth (London, 1997); N. Marinatos, Sir Arthur Evans and Minoan Crete: Creating the Vision of Knossos (London, 2014).
6 For documents discovered to be forgeries in the custody of The National Archives, see <https://discovery.nationalarchives.gov.uk/details/r/C16525> [accessed 29 April 2021].
7 The error was ably exposed in R. Poole, Time’s Alteration: Calendar Reform in Early Modern England (London, 1998), pp. 1–18, 159–78; and Poole, ‘ “Give us our eleven days!” Calendar reform in eighteenth-century England’, Past & Present, cxlix (1995), 95–139.
8 N. Mailer, Marilyn: a Biography (London, 1973, 2012), p. 9. In addition, the term is sometimes used, chiefly in the USA, to denote a trivial fact or ‘factlet’: see <https://en.wikipedia.org/wiki/Factoid> [accessed 29 April 2021].
9 R. Palmer, The Sound of History: Songs and Social Comment (Oxford, 1988).
10 B. Worden, Roundhead Reputations: the English Civil Wars and the Passions of Posterity (London, 2001).
11 A. Prescott, ‘Searching for Dr Johnson: the digitization of the Burney Newspaper Collection’, in Travelling Chronicles: News and Newspapers from the Early Modern Period to the Eighteenth Century, ed. G. Brantzaeg et al. (Leiden, 2018), pp. 51–71.
12 S. Turner, T. Muñoz and P. H. Ros, ‘Measuring mass text digitization quality and usefulness: lessons learned from assessing the OCR accuracy of the British Library’s Nineteenth-Century Online Newspaper Archive’, D-Lib Magazine, xv (July/August 2009): <http://www.dlib.org/dlib/july09/munoz/07munoz.html> [accessed 29 April 2021].
13 S. P. Menafee, Wives for Sale: an Ethnographic Study of British Popular Divorce (Oxford, 1981); E. P. Thompson, ‘The sale of wives’, in his Customs in Common (London, 1991), pp. 404–68; B. Drummond, Frolicksome Women and Troublesome Wives: Wife Selling in England (London, 2018).
14 J. D. Popkin, History, Historians and Autobiography (Chicago, Ill., 2005); N. G. Adamson, Notable Women in World History: a Guide to Recommended Biographies and Autobiographies (London, 1998); J. Burnett (ed.), Annals of Labour: Autobiographies of Working-Class People, 1820–1920 (London, 1974); and many other compilations.
15 A. Ponsonby, English Diaries: a Review of English Diaries from the Sixteenth to the Twentieth Century … (New York, 1923, 1971); J. S. Batts, British Manuscript Diaries of the Nineteenth Century: an Annotated Listing (Totowa, N.J., 1976); H. Blodgett, Centuries of Female Days: Englishwomen’s Private Diaries (Gloucester, 1989); and a magnificent collected set by M. Bird (ed.), The Diary of Mary Hardy, 1773–1809: V vols (Kingston upon Thames, 2013), with accompanying essays in M. Bird, Mary Hardy and Her World, 1773–1809: IV vols (Kingston upon Thames, 2020).
16 See A. Fox and S. Woolf (ed.), The Spoken Word: Oral Culture in Britain, 1500–1850 (Manchester, 2002) and guide to oratory in B. MacArthur (ed.), The Penguin Book of Historic Speeches (London, 1995).
17 On this source, see E. T. Bannet, Empire of Letters: Letter Manuals and Transatlantic Correspondence, 1680–1820 (Cambridge, 2005); S. E. Whyman, The Pen and the People: English Letter Writers, 1660–1800 (Oxford, 2009); and L. Hannan, ‘The imperfect letter-writer: escaping the advice manuals’, in P. J. Corfield and L. Hannan (ed.), Hats Off, Gentlemen! Changing Arts of Communication in the Eighteenth Century/Arts de communiquer au dix-huitième siècle (Paris, 2017), pp. 53–72.
18 J. Rogers, The Second Best Bed: Shakespeare’s Will in a New Light (Westport, Conn., 1993); M. S. Hedges, The Second Best Bed: In Search of Anne Hathaway (Lewes, 2000); G. Greer, Shakespeare’s Wife (London, 2007).
19 For burgeoning interest in ‘the Bard’, see C. LaPorte, The Victorian Cult of Shakespeare: Bardology in the Nineteenth Century (Cambridge, 2020) and also C. Woo, Romantic Actors and Bardolatry: Performing Shakespeare from Garrick to Kean (New York, 2008).
20 S. Shohei. ‘ “Operation Legacy”: Britain’s destruction and concealment of colonial records worldwide’, Journal of Imperial and Commonwealth History, xlv (2017), 697–719.
21 See G. McLennan, Marxism and the Methodologies of History (London, 1981); P. Claus and J. Marriott, History: an Introduction to Theory, Method and Practice (London, 2017); J. W. Moses and T. L. Knutsen (ed.), Ways of Knowing: Competing Methodologies in Social and Political Research (London, 2019).
22 W. Horn, ‘The potential and limitations of radiocarbon dating in the Middle Ages: the art historian’s view’, in Scientific Methods in Medieval Archaeology, ed. R. Berger (London, 1970), pp. 23–87.
23 A distinction first made by the British philosopher Gilbert Ryle (1900–76): see G. Ryle, On Thinking (London, 1979). For assessments of ‘soft’ techniques, see ch. 11.
24 R. Floud, An Introduction to Quantitative Methods for Historians (London, 1973, 1979); P. Hudson, History by Numbers: an Introduction to Quantitative Approaches (London, 2000); C. Feinstein and M. Thomas, Making History Count: a Primer in Quantitative Methods for Historians (Cambridge, 2002).
25 See further reflections in R. W. Fogel, ‘The limits of quantitative methods in history’, American Historical Review, lxxx (1975), 329–50; J. S. Lyons et al. (ed.), Reflections on the Cliometrics Revolution: Conversations with Economic Historians (London, 2008).
26 Hudson, History by Numbers, pp. 53–167.
27 A. Campbell et al., Research Design in Social Work: Qualitative, Quantitative and Mixed Methods (Los Angeles, Calif., 2016).
28 See P. J. Corfield, ‘Problems in classifying social class by occupation’, in Elections in Metropolitan London, 1700–1850, Vol. I, ed. Green, Corfield and Harvey, pp. 441–56.
29 Quotation commonly ascribed to Albert Einstein, as seemingly reflecting his views, though the attribution remains uncertain: see D. Hirshman, ‘Adventures in fact-checking: Einstein quote edition’: <https://asociologist.com/2010/09/04/adventures-in-fact-checking-einstein-quote-edition> [accessed 29 April 2021].