On this page
The foxglove is a tall garden plant with purple bell-shaped flowers. It is also poisonous. In 1785 a doctor in Birmingham published a book on how to use it as a medicine without killing the patient, and the way he wrote that book matters as much as what he found. This post tells William Withering's story from his own book and from historians of medicine, explains why reporting every case was unusual, and works through a simple example of how selective reporting can fool you.
Who was William Withering?
A biographical sketch by Michael Lee in the James Lind Library records that Withering was born in March 1741 in Wellington, Shropshire, the son of a surgeon. He studied medicine in Edinburgh, qualifying in 1766, practised in Stafford, and in 1775 moved to the General Hospital in Birmingham. He was also a serious botanist and chemist: he published a guide to British plants in 1776, studied the ore later named witherite after him, and was elected a Fellow of the Royal Society in 1785. Illness forced him to retire in 1792, and he died in 1799.
Where did the foxglove idea come from?
Withering tells the story himself in An Account of the Foxglove and Some of its Medical Uses, dated "Birmingham, 1st July, 1785". In 1775 he was asked about a family recipe for dropsy, a swelling caused by fluid building up in the body. The recipe "had long been kept a secret by an old woman in Shropshire". It contained twenty or more herbs, but Withering, already an expert botanist, judged that "the active herb could be no other than the Foxglove".
Later retellings turned this into a vivid scene, with paintings of Withering meeting the herbalist. The cardiologist Dennis Krikler, writing in the British Heart Journal in 1985, traced that picture and showed it was an imaginative rendering. Withering's own few lines are the only real source.
Folk use of the plant was not new either. The James Lind Library notes that the herbalist Leonhart Fuchs had named it Digitalis in 1542. What was new was Withering's method.
How did he work out a safe dose?
Withering spent an hour a day giving advice to poor patients, and he writes that between two and three thousand people a year came. That gave him many chances to try the plant, and many chances to get it wrong. He admits that at first he gave doses "very much too large" and kept them going too long.
Over about ten years he made several careful changes:
He used the leaves, not the root, gathered when the plant was flowering, and dried them so their strength was more consistent.
He replaced long boiling with an infusion, and later used the powdered leaf.
He found that the plant's useful effect on urine did not depend on making the patient sick. Nausea and vomiting were warning signs, not part of the cure.
He also describes a travelling tradesman whose wife gave him a strong brew of green foxglove leaves. Withering found him vomiting, with blurred vision and a pulse of forty a minute. "This good woman knew the medicine of her country, but not the dose of it, for her husband narrowly escaped with his life."
Why did reporting every case matter?
In his preface Withering says it would have been easy to pick his best cases, "But Truth and Science would condemn the procedure. I have therefore mentioned every case in which I have prescribed the Foxglove, proper or improper, successful or otherwise." The book then numbers 163 of his own cases, followed by cases sent by other doctors.
He ends with cautious inferences. The first is "That the Digitalis will not universally act as a diuretic." The last is "That it has a power over the motion of the heart, to a degree yet unobserved in any other medicine, and that this power may be converted to salutary ends." The James Lind Library notes he did not make a clear link between the heart, dropsy and fluid build-up. That link came with later medicine.
Worked example: how selective reporting misleads
These are sample numbers for a classroom exercise, not Withering's data.
A doctor treats 50 patients with a new remedy. 20 improve and 30 do not.
If the doctor reports only the 20 who improved, a reader sees 20 out of 20, or 100%.
The honest figure is 20 ÷ 50 = 0.4, or 40%.
Now suppose a second doctor reports every case of an older remedy: 30 improved out of 60, which is 30 ÷ 60 = 50%.
The selective report makes the new remedy look twice as good as the old one (100% against 50%), when the full data say it is worse (40% against 50%). Only the denominator, the number of patients treated, tells you the truth.
What mistakes do students make with data like this?
Ignoring the denominator. "20 patients improved" means nothing until you know how many were treated.
Counting only the cases you remember. In a lab, record every reading, including the ones that look wrong, then decide openly what to exclude.
Treating an inference as a law. Withering wrote "will not universally act". Good conclusions say where they stop.
If you want to practise turning raw counts into fair comparisons, the Fermi Problem Library has a step-by-step solver that explains the reasoning, which helps when a percentage question hides its denominator. For the idea of a fair test taken further, read our piece on Ronald Fisher and the design of experiments.
What can a student take from how Withering worked?
Standardise before you compare. Same part of the plant, same season, dried the same way. In a practical, that means same apparatus, same method, every time.
Report the failures. His willingness to list "improper" and unsuccessful cases is why his book is still respected.
Write conclusions that admit their limits. His first inference was a warning, not a boast.
Next step: in your next practical record, write down every reading you take, including any you decide not to use, with a one-line reason.
