Mind's Eye: How One Ancient Latin Invented Our Way to Visualize Stories

Mind’s Eye

How One Ancient Latin Invented Our Way

To Visualize Stories

Jean-Baptiste Piggin

Cernimus

2018

This edition copyright © 2018-2024 Jean-Baptiste Piggin

After December 31, 2024: CC-BY-SA-4.0

Moral rights reserved

ISBN 978-0-473-45753-2 (e-Book edition)

Cover image by LBM1948 of Wikipedia.

Plates 1 and 2 by Biblioteca Medicea Laurenziana, Florence.

All other art by the author.

Published 2018 by Cernimus

PO Box 6456, Auckland 1141, New Zealand

Version 1.02, 2018-11

Visit www.Piggin.net

Twitter: follow @JBPiggin

Subscribe to MacroTypography.Blogspot.com

for

Sabine

Cover: Hormosira banksii growing in a tidal pool. This kelp, found only in New Zealand and southeastern Australia, is one of the few plant species to grow in stemmatic fashion, forming nodes and connectors and branching from the nodes. Colorized from a photograph by LBM1948 for Wikipedia. This image: CC-BY-SA-4.0

Chapters

Figures

[→ 1]

Europe’s oldest extant diagrammatic genealogy of a historically documented family dates from about 1013

[→ 2]

How we misread stemmata: we think the root must dominate, but the main subject may often be a remote branch

[→ 3]

Like a tablet computer, a good stemma is a visual-tactile medium suited to the size of fingers and the visual field

[→ 4]

Genealogy is only one among the uses of a stemma - all categorical thinking can be enhanced by visualization

[→ 5]

Stemmata are equally useful to analyse processes: as flow charts, they make chains of decisions more understandable

[→ 6]

A first finding: mapping hierarchical coordinates to a stemma makes them more easily intelligible to human readers

[→ 7]

The search for the most primitive stemma: Europe’s oldest genealogical drawings begin with an image of Adam and Eve

[→ 8]

The essential feature that defines a stemma: one or more child nodes attach themselves to a parent node

[→ 9]

The tables are an early form of family tree: the layout uses roundel shapes and mixed directions of entrainment

[→ 10]

A first riddle in the Spanish tables: how to understand the logic of its sudden swerves in direction

[→ 11]

Oddities in the Spanish tables include horizontal chains of roundels running parallel to one another across the page

[→ 12]

A second riddle of the tables: why are roundels docked to a bus bar like garments hung on a clothesline?

[→ 13]

A third riddle of the tables: how to interpret pendant chains with the siblings attached to one another vertically

[→ 14]

A conceivable improvement to the Spanish tables’ pendant chains would have been to draw fans in ray form

[→ 15]

Another alternative to the pendant chain would be to draw a vertical bus bar denoting a shared parental connection

[→ 16]

A curiosity: the mysterious clothesline style could be read as a bus bar, though its purpose is unknown

[→ 17]

Depicting polygamy: an ingenious layout masters this genealogical-drawing challenge in the tables

[→ 18]

The pages of the Spanish tables are not self-contained but visibly connect to form larger graphic assemblies

[→ 19]

Wilhelm Neuss postulated that the biblical genealogies were organized into four distinct graphic units or tables

[→ 20]

Yolanta Załuska distinguished six different versions of the genealogical tables in her list of surviving manuscripts

[→ 21]

In every manuscript, many of the kings of Judah are allotted wives whose names are given in no other ancient source

[→ 22]

The mismatched couples are a transmission error, indicating that all our versions of the tables have been corrupted

[→ 23]

The Neuss partition turns out mistaken - the genealogies, hereafter the Great Stemma, were created as a single document

[→ 24]

The Great Stemma carries no trace of Jerome’s Vulgate Bible: it relies on an earlier Latin text of the Old Testament

[→ 25]

Might the Great Stemma have been first drawn as a chart to decorate a wall or as a souvenir? Doubtful

[→ 26]

Or might the Great Stemma have been created to assist students learning every name in the Bible? Also doubtful

[→ 27]

The least corrupted manuscript of the Great Stemma is located in Florence: its treatment of Jesus distinguishes it

[→ 28]

To reconcile a contradiction between Matthew and Luke, the Great Stemma attached one of the two ancestries to Mary

[→ 29]

A rival view of the conflict, propounded by Julius Africanus, had applied both of the gospel ancestries to Joseph

[→ 30]

The Julius Africanus solution depended on an unusual constellation of marriages and births over three generations

[→ 31]

The Great Stemma solution inserts a man named Joachim, unknown to the canonical gospels, as father of Mary

[→ 32]

The roundels docked in the so-called clothesline style along the top margins may reward closer attention

[→ 33]

Modern readers would expect to see an extended father-son sequence depicted like beads on a necklace

[→ 34]

Also notable in the Florence manuscript: the clothesline style exists on the bottom margin too, but inverted

[→ 35]

The peculiar bus docking style of the Florence manuscript is consistently applied right across the chart

[→ 36]

Testing the docking theory: non-ancestors of Jesus are only shown necklace-connected, interrupting the bar docking

[→ 37]

Analysis shows the whole genealogy of Jesus is reserved to the upper and lower margins of the Great Stemma

[→ 38]

None of the Cainite family connects to a bus bar: conclusive proof that a bus is the mark of Jesus’s ancestral lines

[→ 39]

Expositor’s bus foreshadows an idea in electrical drafting: parallel connections distribute potential equally

[→ 40]

The ancestries at the margins, hereafter fila, are of prime importance among the Great Stemma’s graphic features

[→ 41]

The Great Stemma is equivalent to a flow-chart, demonstrating an idiosyncratic interpretation of biblical history

[→ 42]

Building the Great Stemma was an engineering task tackled bottom-up - small assemblies had to be designed first

[→ 43]

Visualizing a kingly Israelite household: position the children below, the father above, and the harem in between

[→ 44]

The irregular cases – childless or multi-child unions – can be appended to the harem’s core visual structure

[→ 45]

Advancing the replay: each ostracon becomes a roundel, and linking lines gather these into assemblies

[→ 46]

Thanks to figure-ground perception, small connectors stand out for human readers better than large connectors

[→ 47]

Psychology for diagrams: Identifying is self aware, but may be weak; perception of location is non-aware, but strong

[→ 48]

The dorsal-ventral hypothesis endeavours to explain why our perceptions about where are largely unconscious

[→ 49]

Rule-driven positioning: an invisible hallmark of infographics catering well to the foibles of human perception

[→ 50]

Size changes are natural ways to create a texture in the data content in node-link diagrams and in displays

[→ 51]

Big roundels cater to our perceptual tendency to favour large items provided they are rare in our field of view

[→ 52]

Reprising a definitive feature: in the sub-stemmata, a fan of rays often begins the ramifications away from the root

[→ 53]

The five-generation sub-stemma of Rachel is dense with pendant chains: a good test case to investigate their purpose

[→ 54]

Hierarchies, which are usually hard to think about, become easier to grasp if we think of them as navigation issues

[→ 55]

Because hierarchical processing is difficult, children initially find stemma drawing non-intuitive and confusing

[→ 56]

Why straight-line chains are beneficial: in life and on paper, humans naturally explore the directest paths first

[→ 57]

A ray-form method to present the Rachel sub-stemma can be imagined, but was never employed: has it disadvantages?

[→ 58]

Decay: Designed to occupy a fixed height of 10 rows of roundels, the Great Stemma was corrupted by repeated copying

[→ 59]

Pendant chains persist in a new guise in early modern European genealogies: as name-lists incorporated into stemmata

[→ 60]

Spatial reasoning requires no orientation: the Great Stemma emphasizes this by inverting part of its text

[→ 61]

A general principle: stemmata remain fully legible no matter where the root is placed in our frame of view

[→ 62]

Groups of chronographic data in the Great Stemma provide a political history in a highly compressed form

[→ 63]

Expositor’s time sequences are not all straight timelines: some are curves: they can be wayward, tangled and snipped

[→ 64]

Another investigation: why is Pentateuch material omitted from the twin tracks of the kings of Judah and Samaria?

[→ 65]

On the way to a solution: the Great Stemma omits certain Judaean kings, yet names their wives: Why?

[→ 66]

For each of the seven missing reigns, some kind of evidence of its former inclusion can be found in the wreckage

[→ 67]

Placement of the missing reigns: they would have been wedged between Matthew’s kings, without docking to the top bar

[→ 68]

The twin tracks would have been roughly equal in length, graphically emphasizing their parallel nature

[→ 69]

The docking shapes in the Great Stemma encode specific types of relationship: a summary

[→ 70]

Justus, a Spanish notary who died in 772, evidently saw at least part of the Great Stemma in a book

[→ 71]

Even if Justus only saw a chad of it, there can be no doubt the Great Stemma existed in his lifetime

[→ 72]

A brief chronicle which supplements the Great Stemma, the Ordo Annorum Mundi, demonstrably existed by 636

[→ 73]

Tracing the early history of the western node-link diagram relies on these key documents

[→ 74]

To investigate the Liber Genealogus of 427, we begin by sampling the name order in the Septuagint Genesis

[→ 75]

In the Liber Genealogus, the descendants of Keturah are set out in an order different to that of Genesis

[→ 76]

The meaning of breadth-first traverse: process all of each generation fully before proceeding to a next generation

[→ 77]

By contrast, a depth-first traverse prioritizes ramifications, processing nodes towards the outer extremities first

[→ 78]

(1) The natural depth-first reading order of the stemmata is reproduced in the Liber Genealogus‘s unbiblical order

[→ 79]

(2) The chart’s left-to-right, top-down order show shows up again in the Liber Genealogus chapter order

[→ 80]

(3) Even the repetitions necessitated by the Great Stemma’s organization are echoed in the Liber Genealogus text

[→ 81]

Dating the Latin works with respect to one another provides the proof that Expositor was alive in the 4th century

[→ 82]

Data positioned off-beam attracts attention because our vision is programmed to note exceptions (outlier technique)

[→ 83]

Outliers in modern graphs are a refinement of an attention-tuning technique Expositor that was already using

[→ 84]

By visualizing logical problems, many people can solve them without having to translate them into propositions

[→ 85]

To resolve two very similar items visually, one brings them into direct proximity (contrast technique)

[→ 86]

Visual analytics assumes that connections within data are more easily understood when graphed than when narrated

[→ 87]

To portray an encounter, the visualizer uses proximity to suggest a nexus, exploiting perceptual clumping

[→ 88]

Careless of shape and positioning, diagrams for Greek geometry and science failed to fully exploit visualization

[→ 89]

Because it demands attention to the bead positions, the abacus makes effective use of visualization

[→ 90]

Apart from the Great Stemma, clear early evidence for tiny stemmata comes from a definition by Isidore of Seville

[→ 91]

A mention of stemmata in the Pauli Sententiae (about 330 CE) apparently prescribes a method to decode them

[→ 92]

Cassiodorus (c 562) drew stemmata of ideas and used the Latin term linealis descriptio for a visualization

[→ 93]

Even before the rise of the stemma, the Romans were developing a visualization technique with the arbor juris

[→ 94]

The most ancient arbor juris dates from the 2nd century CE or earlier and resembles a scaffolding of ladders

[→ 95]

Game boards storing real-time positions in gambling races were the likely original impetus for data visualization

[→ 96]

Fila C and D of the Great Stemma follow curves similar to a Game of Fifty-Eight Holes board: Is this a coincidence?

[→ 97]

Partitioning the data: the Great Stemma gathers a portion of its chronological data into panels shaped as arches

[→ 98]

An excess of information forces a separation: Genesis provides 4 facts about every patriarch prior to Abraham

[→ 99]

The multiple-view solution: how Expositor cached all the vital data on the 19 patriarchs in arcades

[→ 100]

A modern solution: how graphing would allow us to visualize the relative lifespans of patriarchs over 3,500 years

[→ 101]

In the 16th century, stemmata were re-invented as family trees with bottom roots, trunks and leaves

[→ 102]

A precursor to the family tree: the 12th-century Virga Jesse loosely associated Jesus’s ancestry with a living plant

[→ 103]

The 12th-century fondness for the tree motif led artists to sometimes impose tree decoration on the arbor juris

[→ 104]

In 12th-century analytical teaching, trees served as a canonical visual form when memorizing seven-part collections

[→ 105]

The earliest family trees in a true sense: royal portraits and emblems arranged on vines by Renaissance engravers

[→ 106]

How trees mislead: the model blinds us to the fact that branches which do not extinguish may ultimately recombine

[→ 107]

The graphic disadvantage of trees: trimming the twigs for aesthetic reasons obscures or falsifies data structures

[→ 108]

How the Great Stemma sank from sight: Peter of Poitiers’ history chart, the Compendium, made it obsolete

[→ 109]

The modern era: with the advent of printing, the roundels of stemmata were often replaced by rectangular clipei

[→ 110]

Instead of connecting lines, printed stemmata employed braces to connect multiple offspring to parents

[→ 111]

The fila re-appear in a Chinese Christian poster of 1704 which explains one view of the genealogy of Jesus

[→ 112]

A grand finale: the Tú of Shandong offers a modern-era interpretation of Expositor’s collision zone

Plates

[→ 1]

The opening of the Great Stemma in Florence, beginning with Adam and an arch of patriarchs

[→ 2]

The final page of the Great Stemma in Florence, ending at Jesus with a mention of King Wamba

[→ 3]

A clean reconstruction of the original Great Stemma, with text translated into English

People in the Story

Marcus Tullius Cicero (106 – 43 BCE), Roman orator

Coluccio Salutati (1331 – 1406), Italian humanist

Cunigunde of Luxembourg (c. 975 – 1040), queen of Germany and Holy Roman empress

Eusebius of Caesarea, Hellenic chronographer (260/265 – 339/340)

Expositor, a 21st-century pseudonym for the anonymous author of the Great Stemma

Gregory the Great (c. 540 – 604), philosopher and pope

Christophe Guignard, Swiss scholar and editor of Revue des Sciences Religieuses

Henry II (973 – 1024), Holy Roman Emperor, married Cunigunde

Henry III (1017 – 1056), Holy Roman Emperor, married Agnes

Hugh of St Victor (c. 1096 – 1141), Paris-based philosopher

Jerome of Stridon (c. 347 – 420), Latin translator of the Bible

Jesus of Nazareth (c. 4 BCE – 30/33), founder of Christinanity

Julius Africanus (c. 160 – c. 240), Hellenic controversialist

Justus (c. 710 – 772), a notary in Spain

Christiane Klapisch-Zuber, historian of family trees

Markus Knauff, German cognitive theorist

Wolfgang Lazius (1514 – 1565), Austrian historian of the Hapsburg empire,

Luke the Evangelist, 1st-century author of one of four canonical Christian gospels

Maius or Magius (c. 900 – 968), artist in Leonese kingdom

José Carlos Martín Iglesias, Spanish philologist

Matthew the Evangelist, 1st-century author of one of four canonical Christian gospels

John Pierpont Morgan (1837 – 1913) and John Pierpont Jack Morgan jr. (1867 – 1943), father-and-son banking tycoons and founders of a New York library

Wilhelm Neuss (1880 – 1965), German art historian

J. Kevin O’Regan, US researcher in Paris into visual perception

Nathaniel Lane Taylor, editor and publisher of The American Genealogist

Yolanta Załuska, Polish-born art historian in Paris

1: A Whole Dynasty at a Glance

With her husband Henry, Cunigunde was picking her way southwards over the Alps. Also along for this work-related trip in November 1013 was the couple’s army and baggage. And possibly a slip of parchment with a little sketch on it.

For an alpine crossing, November is late in the season and perilous, even for a queen with a horde of wild-haired German warriors sworn to ensure her safety. As winter approaches, the snowline is descending and the hours of daylight are growing short. The armed host was conducting a two-month crossing from Germany to Pavia, capital of the Lombard kingdom of northern Italy. The summit of the Brenner Pass was less of an impediment to the king, queen and courtiers than a long gorge further south where the Eisack River narrows to white-water rapids between towering rock walls.

At the gorge entrance, the alpine guides would have advised Henry to take the same route he had used nine years earlier, when he visited Italy to put down a revolt by massacring part of the population of Pavia. The old Roman waterside road through the base of the gorge had almost certainly become impassable from centuries of neglect, with its ingeniously engineered bridges washed out and its narrow ledges blocked by rock falls. A next-best route required an exhausting 850-meter ascent, winding through thickly wooded ravines to a plateau almost as high as the Brenner, followed by the vertigo of a slithering 900-meter descent on an even steeper muddy track to the alluvial plain of the Adige river. For humans and horses this sally into the high wilderness was bad enough; for the baggage train, which probably employed carts, it would have been a vast, curse-filled organizational confusion.

A thousand years later, I am standing at the spot, wondering what documents Cunigunde carried in her travel trunks as she rode past.

My eyes have gone wide-angle from the grandeur of the mountain scenery: a vast, block of dolomite, the Schlern, flanked by jagged pinnacles, rises 2,000 meters above the deep slit cut by the Eisack.

To a medieval German queen, there would have been no delight in this journey at all. The view would have been simply fearsome, and the terrors were heightened by the risks of ambush, of falling, of death in the cold. She would not have undertaken the odyssey at all if she had not been burning with a supreme ambition nursed over the past twelve years: to become empress of the West. The proof is a diagram that lays out in graphic form her claim to this ultimate earthly rank. It demonstrated that Cunigunde was descended from Charlemagne, who had seized modern Germany, France and northern Italy to found a Frankish empire in 800 CE.

Charlemagne had claimed for his domains the dignity of a new Roman empire, a western equal to the rump Roman empire of the Byzantines ruled from Constantinople. Its public administration and religion were conducted in Latin, though its ruling elite spoke a dialect of German. Charlemagne’s empire soon fell apart, but this one female descendant two centuries later still held his ambitions dear. The drawing, by now the oldest surviving genealogical diagram in Europe – perhaps anywhere – to depict a non-mythical family, tells the reader that Charlemagne’s noble blood now courses in a woman’s veins.¹ She was on her way to meet the only person with God’s authority to crown her as empress: the pope in Rome.

Europe’s oldest extant diagrammatic genealogy of an historically documented family dates from about 1013:

Ancestors of Cunigunde (C) included Charlemagne (R), an emperor, and Arnulph (A), a saint. Her husband Henry (H) from an unrelated family was a distant cousin to the three Ottos, I, II and III (right). Source: München, BSB, clm 29880(6.

My digital tracing of Cunigunde’s diagram omits the Latin text (since it would be illegible at this scale) but I have labelled the most important of the 42 circles. C marks our queen with her forthcoming title, Cynigund imperatrix in Latin, in the circle at the bottom left. Her forebears and her French cousins are placed at the centre of the diagram in a vertical line that resembles a totem pole, topped by their ancestor, Arnulph (A), who had abandoned his children for a life of prayer and was recalled to public service in 612 as bishop of Metz.²

Halfway down the pole is royal Charlemagne (R), whose empire was split in three by his squabbling grandsons under the Treaty of Verdun in 843 to settle a three-year war.

At the left, hitched up like a festoon, is a group which comprises the rulers of Italy-cum-Lorraine, one of the three realms created by that split, plus their distaff descendants. Another bracket leads to a festoon at right comprising the branch of Charlemagne’s offspring which ruled East Francia – essentially Germany – after that break-up and then died out. The design is not only elegant, but also functional, showing how the people involved in the break-up were inter-related.

Detached at the bottom right is a sparing little family chart of Cunigunde’s husband Henry (H), who was a second cousin of the three Ottos (O1-O3). The Ottonian emperors were German parvenus, who had seized and partly rebuilt the empire, but were unable to claim blood connections to the venerated Charlemagne.

The implication shining through the diagram is plain: he might be a full-bearded alpha male, but Henry was of inferior stock. Cunigunde alone was a member of an old imperial family, surpassing anything crafty upstart Henry or even the downmarket Ottos could claim. But with help from his wife, Henry could partake of the deep-down imperial dignity they both craved.

This is not a family tree. Not until many, many years after this parchment was made, were such diagrams rebranded as family trees and redesigned to resemble oaks. In late Renaissance Italy, such diagrams would be enhanced with bark-covered trunks, green leaves and fruit and the once simple connecting lines would morph into woody branches to fulfill the new tree metaphor. Tree drawing captured the imagination of all Europe ever after. Today we find it difficult to not think, That’s a tree, when we see such a drawing.

piggin.net/cn.htm

to: BSB library

In Cunigunde’s day and earlier, there was no foreknowledge of that future style craze. This German precursor of trees was just a back-office technical diagram without any arboreal decor. The drawing – perhaps not Cunigunde’s original, but a copy – is preserved in the Bavarian State Library (BSB) in Munich, Germany: a tattered page measuring 19.3 by 14.8 centimeters found bound into a book at a Regensburg monastery.

An old parchment, especially one in an unfamiliar language, has the monstrosity of an aged stranger’s wrinkled face, yielding up no easy insights about happier younger days. If you are able to overlook its current ugliness, discern its long-faded loveliness and sense its power to enthrall, you can find online each of the manuscripts I will be mentioning. But that is not obligatory since we can recreate all of these documents in the freshness of their first days. My plots, traced on a computer screen using vector-drawing software, reproduce the essentials of the drawings – their form – removing distractions such as decorations, doodles, graffiti, fading, buckling and tearing. You will see only the clean outlines the artist first drew on creamy white animal skin.

The Munich library catalog terms the document the Tabula genealogica Carolingorum – a genealogical infographic of the Carolingians – as if one ought to begin to reading it at Charlemagne at the middle and as if the diagram’s purpose were to explain his kingly line. But that is not what this diagram was drawn for. Nor is it the descendants of Arnulph, another occasional title, as if it should be read from its top by a reader curious about how many offspring a non-celibate saint can generate. There can be no doubt that the person this diagram is about is neither Arnulph nor Charlemagne, but Cunigunde. A proper title ought therefore to be something like the ancestry of Cunigunde in infographic form. Or the Stemma of Cunigunde, since the word stemma, anciently a genealogy in the broad sense, had become a technical term in medieval Latin for ramifying drawings like this.

How we misread stemmata: we think the root must dominate, but the main subject may often be a remote branch:

The true reading of the Stemma of Cunigunde begins at the bottom edge, at Cynigund imperatrix (Empress Cunigunde), and proceeds via her father Count Siegfried, his mother Cunigunde and his grandmother Ermentrude to the main line at Hludowicus, King Louis the Stammerer, who ruled 877-879. Below him are the Latin names of three of his West Frankish descendants: Charles III (ruled 898-922), Louis IV (936-954) and Lothair (954-986). ? marks a roundel now illegible which denoted Louis V.

The impressed reader was meant to start reading this diagram at the bottom, at Cunigunde’s circle, tracing her bloodline back to join the ramrod-straight main line of the rulers of the kingdom of West Francia. The death in a hunting accident in 987 of childless King Louis V (his name on the bottom margin is no longer legible on the abraded parchment) had brought the line of male Carolingians to an end. Had it not been for Cunigunde. Twenty-seven years later, she had business to do in Rome, stepping up to assert the renewed vitality of Charlemagne’s blood.

The Stemma of Cunigunde is a good place to start investigating why human beings visualize data. Its quaintness and its odd conventions serve better than any modern diagram would to highlight the intrinsic strangeness of diagramming. This lets us analyse it – and the much earlier chart which will be the principal subject of this book – free of most preconceptions.

My method is historical, but my purpose is not. To ascertain what drives the human capability to visualize data, we need to compare our own visualizing culture with some sophisticated human society that does not visualize, and then observe what would change if diagrams and visual interfaces were to be adopted by adults of the non-visualizing culture. It might still be possible to monitor such a process happening in the world today, but that would require costly field research. Fortunately, we have access at almost no cost to historical data which documents this transition as it took place in western Europe. We seldom reflect on why we often find visualizations easier to absorb than verbal explanations, nor on what this tells us about the jerry-built, kludged-together nature of human rationality. My observations will come from the past, but the object of this book is to illuminate a capability of the modern human mind.

The blobby style of the Cunigunde diagram may appear exotic to