One Bone, One Source: Checking Bones, Trees and Fossils
In the drawings on this site, one bone always carries one colour. The written claims follow a similar rule: each statement about a bone, a fossil, a family tree or an animal's building habits leads back to a source a reader can follow. This page sets out where those sources come from, how the figures are drawn from them, and what happens when sources disagree.
Which sources count, and in what order
When sources of different kinds disagree, the higher rank carries the claim:
- Original research papers. These have the last word on anatomy and on how animals are related. Links go to the DOI or to the paper's own open page wherever one exists.
- Specialist data. The Open Tree of Life supplies tree topology. NCBI Taxonomy is used only to match names, spellings and identifiers, never as evidence of how groups are related, since NCBI itself does not present its classification as a primary source on relationships.
- Institutional pages, such as those of the U.S. National Park Service, the U.S. Fish and Wildlife Service and NatureScot. They support facts; they do not settle questions of anatomy or relationship.
Public-domain classics, from Richard Owen's 1843 lectures and Charles Darwin's On the Origin of Species (1859) to Thomas Henry Huxley's papers on birds and reptiles, are quoted for what they said in their own day, with the year attached. Owen's 1843 glossary, for example, defines a homologue as "the same organ in different animals under every variety of form and function." When an older idea has been overturned, the page gives the source where it first appeared and the source that overturned it, as a pair.
General encyclopedias and wikis can point toward a source. They are never cited as one.
When two sources of the same rank disagree, both are given side by side and neither is quietly dropped. Whether the single fossil feather described in 1861 belongs to Archaeopteryx, for instance, is argued in opposite directions by studies published in 2019 and 2020, and a page on it reports both.
What always needs a source
- any claim about how a particular species behaves, which covers every page on animal builders;
- where a group sits on the tree, and how two groups are related;
- the age of a fossil and the way it was preserved;
- the species name attached to a fossil, specimen or photograph;
- sizes, counts and other numbers.
How much of a source was read
Each entry in a page's source list says whether the full text was read or only an abstract or summary. Claims that rest on an abstract keep the abstract's own level of caution. A work known only from a bibliography can be listed for readers who want to follow it, but nothing in a title, the opening answer or a page summary may rest on it. Papers still under copyright are paraphrased, and quoted only in brief.
Same bone, same colour
Every bone that belongs to the shared forelimb, jaw, ear or trunk plan has one colour, fixed for the whole site and never chosen page by page:
| Bone | Colour | Contrast on the plate |
|---|---|---|
| humerus | #275290 | 7.02:1 |
| radius and ulna | #876473 | 4.62:1 |
| carpals | #5A6236 | 5.85:1 |
| metacarpals | #007873 | 4.81:1 |
| phalanges | #A3501C | 5.09:1 |
| dentary | #736E69 | 4.55:1 |
| articular, the same bone as the malleus | #5F2864 | 9.59:1 |
| quadrate, the same bone as the incus | #377D46 | 4.52:1 |
| squamosal | #282D69 | 11.34:1 |
| stapes | #825FA5 | 4.58:1 |
| vertebrae | #644614 | 7.80:1 |
| ribs | #0078A0 | 4.51:1 |
| pelvis | #00505F | 8.18:1 |
The colours were chosen so that bones drawn next to each other stay distinct in normal colour vision and in simulations of the three common forms of colour-vision deficiency, using the simulation model of Machado, Oliveira and Fernandes (2009). Each neighbouring pair is at least 10 apart on the CIEDE2000 colour-difference scale under all four conditions, and every colour has a contrast of at least 4.5 to 1 against the drawing's background.
Colour is the second cue, never the only one. Each figure sits on a fixed bone-white plate in both light and dark display modes, every bone has a thin dark outline, and every bone carries its name as a label. A bone outside the table, such as a wishbone or an ankle bone, is drawn as an outline with a label and no fill. Bones that fuse in some animals, such as the carpometacarpus of a bird's wing, are drawn as segments in the colours of the bones they contain, and the legend says this is a drawing convention, not a statement about anatomy.
Counts and measurements
Bone counts and sizes come from published studies and open data. Wherever a table can be downloaded, each row stands on its own: it names its source and carries a qualifier column for ranges, approximate values and alternative readings. Only sources released under open licences or in the public domain go into a downloadable table. Figures from copyrighted papers, or from sources read only as an abstract, are paraphrased in the text instead, and the matching row of a figure is marked "not compiled". Unit conversions and everyday size comparisons are recalculated; recalculation confirms the arithmetic and the presentation, while the input numbers rest on their sources.
Family trees
- Where the shape comes from. A tree of living animals is redrawn from the Open Tree of Life synthetic tree, and the figure names the synthesis version and the date the data were retrieved. A tree that includes fossil groups is built branch point by branch point from published phylogenetic studies, and each branch point names the study that supports it. The layout of a paper's own figure is never copied.
- No time scale. Branches are drawn without lengths and without dates, and the figure says so. The left-to-right position of a split shows only its order.
- Weak branches look weak. A grouping supported only by taxonomy, with no phylogenetic study behind it, is drawn as a dashed line and explained in the legend. A split that the data leave unresolved is drawn as several branches from one point.
- Names must match exactly. Each name is matched exactly; near matches and suggested spellings are rejected. If a group has been dropped from the synthesis, or the tree returned does not hold the groups that were asked for, that tree is not drawn from those data.
- Turning the branches. Every tree is redrawn with its branches flipped and shuffled, and the groupings are compared. They must come out identical, because turning a branch point never changes a relationship.
- Ancestors are marked. On a rooted tree, each inner branch point stands for a common ancestor. A tree without a root is labelled unrooted, and no direction of descent is drawn on it.
Trees are pulled again and redrawn every twelve months, since placements move as new studies appear.
Fossils and the deep past
Every placement, date or claim about how an extinct animal looked names the study and its year. When accounts differ, both stand on the page: reports of when the Anning family's first ichthyosaur skull came out of the cliff differ by a year, and pages give both. The colour of an extinct animal is given only as an inference, with the study that drew it. Statements about fossils are checked again against the original literature every twelve months, and numbers that change as finds are made, such as how many specimens of a species are known, carry an "as of" year.
Photographs and plates
Every image is checked on its own. Its licence and licence version are read from the file's own record, and the reputation of the museum or archive that holds it counts for nothing on its own. Four kinds of terms are accepted: CC0 dedications, Public Domain Marks, and the CC BY licence with its ShareAlike variant, CC BY-SA. Images with non-commercial or no-derivatives terms are not used.
A public-domain plate needs a stated reason: the creator's death year and the law that applies, or an explicit public-domain statement from the source. Publication in the nineteenth century is not enough by itself, and a plate whose artist is unknown is not presumed free. For anonymous works, the year of publication decides. Scans that carry their own conditions of use are left out even when the original plate is old.
Works made by U.S. federal employees as part of their jobs, such as photographs by the National Park Service or the U.S. Fish and Wildlife Service, have no copyright in the United States under 17 U.S.C. §105. European law has no matching rule, so outside the United States the basis is the agency's own public-domain marking. These photographs are used with the agency and the photographer named.
A species name appears on a photograph only when the file's record names the species and the picture matches the description. Photographs of look-alike species without an authoritative label are not used. A photograph may be cropped, adjusted in colour or overlaid with bone labels, and the credit line lists each such change; its original background is never cut away.
Illustrations
Generated illustrations are used only for living animals in their surroundings, and each is marked Illustration. Fossils, specimens and extinct species are never generated, and no life reconstruction of an extinct animal appears on the site in any form. Rulers, colour keys, bone charts, trees and cross-sections are always drawn from data.
How to reuse the site's figures and tables is set out on Plates for classrooms. The vocabulary used in the trees is defined at node blindness, and errors can be reported through Tag a question.