Ink patterns are direct records of muscle movement, unique to each person.
Há dois mil anos, mãos anônimas copiaram as escrituras em pergaminhos às margens do Mar Morto — e por décadas, os estudiosos debateram quantas mãos seriam essas. Agora, pesquisadores da Universidade de Groningen utilizaram inteligência artificial para responder a essa questão com precisão estatística: o Grande Rolo de Isaías foi escrito por dois escribas distintos, cujas caligrafia tão semelhantes sugerem uma origem ou formação comum. É um momento em que a tecnologia do presente ilumina o silêncio do passado, transformando a especulação em evidência e devolvendo, de certa forma, a identidade a quem copiou o sagrado.
- Uma questão que persistia sem resposta definitiva por décadas — quem escreveu o Rolo de Isaías? — foi finalmente resolvida com o auxílio de redes neurais artificiais.
- A uniformidade quase perfeita da caligrafia havia frustrado gerações de paleógrafos, tornando impossível distinguir a olho nu onde uma mão terminava e outra começava.
- A equipa desenvolveu algoritmos capazes de isolar traços de tinta em couro milenar e medir padrões musculares únicos de cada escriba, transformando marcas físicas em dados comparáveis.
- A divisão revelou-se clara e estatisticamente significativa: dois grupos distintos de colunas, separados aproximadamente na coluna 27, com o segundo escriba exibindo maior variação nos seus traços.
- O método abre caminho para uma revolução na análise de todos os Manuscritos do Mar Morto, prometendo identificar escribas individuais em múltiplos documentos com rigor científico.
O Grande Rolo de Isaías, um dos manuscritos bíblicos mais antigos alguma vez descobertos, foi escrito por duas pessoas diferentes. Pesquisadores da Universidade de Groningen, liderados por Mladen Popovic e pelo cientista informático Lambert Schomaker, publicaram na revista PLOS One a confirmação computacional daquilo que os estudiosos suspeitavam há anos sem conseguir provar.
O desafio era precisamente a semelhança entre as duas caligrafia. Ao olho humano, os 54 colunas do pergaminho pareciam obra de uma única mão — consistentes demais para revelar uma transição. Para ultrapassar essa barreira, a equipa criou algoritmos que separaram a tinta do suporte em couro e papiro, e desenvolveram uma rede neural artificial capaz de preservar e analisar os traços originais de tinta deixados há mais de dois milénios. Esses traços são registos diretos do movimento muscular, únicos a cada pessoa.
A análise combinou características texturais — como a curvatura dos traços individuais — com características alográficas, que examinam as formas completas das letras. O resultado foi inequívoco: as 54 colunas dividem-se naturalmente em dois grupos, com a transição a ocorrer por volta da coluna 27. Uma análise adicional centrada na letra aleph nas primeiras e nas últimas 27 colunas confirmou diferenças visíveis e mensuráveis. O segundo escriba revelou maior variação na sua escrita, embora os dois estilos fossem notavelmente próximos — indício de que poderão ter partilhado a mesma formação ou tradição escribal.
A importância deste trabalho vai além de um único texto antigo. Os investigadores acreditam ter aberto um novo caminho para o estudo dos Manuscritos do Mar Morto no seu conjunto, transformando a paleografia de uma arte subjetiva numa ciência quantificável. Pela primeira vez, as mãos que copiaram as escrituras há dois mil anos podem ser distinguidas não por argumento, mas por prova.
The Great Isaiah Scroll, one of the oldest biblical manuscripts ever found, was written by two different people. Researchers at the University of Groningen in the Netherlands used artificial intelligence to prove what scholars had long suspected but never quite confirmed: the 2,000-year-old parchment bears the hand of not one scribe, but two, working in sequence across its 54 columns of text.
The discovery emerged from a study published Wednesday in the journal PLOS One, led by Mladen Popovic and his team, which included Lambert Schomaker, a computer scientist specializing in how machines can read handwriting. The Isaiah Scroll was discovered seven decades ago in the caves near the Dead Sea, part of a vast cache of ancient texts that includes the oldest known copies of the Hebrew Bible. What made this particular manuscript interesting to researchers was that its writing appeared almost uniform to the naked eye—so consistent, in fact, that the possibility of multiple hands had been debated among experts for years without resolution.
The team's approach was methodical and computational. They began by creating an algorithm to separate the ink from its background of leather and papyrus, then developed an artificial neural network capable of preserving the original ink traces left by the scribe's hand more than two millennia ago. This step mattered enormously because ink patterns are direct records of muscle movement, unique to each person. The researchers then analyzed two types of handwriting characteristics: textural features, which measure things like the curvature of individual strokes, and allographic features, which examine complete letter forms.
The numbers told a clear story. When the team examined the 54 columns of text, they found they naturally divided into two distinct groups. These groups were not scattered randomly across the parchment but clustered together, with the transition occurring roughly halfway through the document, around column 27. A second analysis using letter-fragment patterns confirmed what the first had suggested: two different people had written this scroll. The second scribe, notably, showed more variation in his handwriting than the first, even though their styles were remarkably similar—suggesting they may have trained together or come from the same scribal tradition.
To further verify the finding, the researchers conducted a third analysis, this time focusing on the letter aleph in the first 27 columns and the last 27 columns. The differences were visible and measurable. Some scholars had previously pointed to physical aspects of the parchment and text placement to argue that a new hand had taken over after column 27, but the theory had never achieved consensus. Now, with statistical rigor behind it, the evidence was undeniable. Popovic explained the significance: instead of relying on impressionistic judgment, the team could demonstrate with computational precision that the division was statistically significant.
What makes this work important extends far beyond a single ancient text. The researchers believe they have opened a new pathway for studying the Dead Sea Scrolls as a whole. By analyzing the physical characteristics of handwriting rather than relying on subjective observation, scholars may now be able to identify individual scribes across multiple manuscripts and understand how each one worked. The method could transform paleography from an art into a quantifiable science, revealing not just who wrote what, but how scribes approached their labor, where they paused, how their hands moved, and what their individual quirks and habits were. For the first time, the hands that copied scripture two thousand years ago can be distinguished not by argument but by proof.
Notable Quotes
Instead of relying on impressionistic judgment, the team could demonstrate with computational precision that the division was statistically significant.— Mladen Popovic, University of Groningen
The second scribe presented more variation in his writing than the first, although the two writings were very similar.— Lambert Schomaker, computer scientist
The Hearth Conversation Another angle on the story
Why does it matter that two people wrote this one scroll? Couldn't one person's handwriting just change over time?
Handwriting does change, but not in the way we see here. The key is that the two sections are clustered—one scribe did columns 1 through 27, the other did 28 through 54. If it were one person's hand evolving, you'd expect the shift to be gradual, scattered throughout. Instead, there's a clean break.
And the AI could see that when human eyes couldn't?
Not exactly couldn't—some scholars suspected it. But suspicion isn't proof. The computer measures thousands of micro-details in how ink curves, how letters connect, the pressure patterns. It finds statistical significance where the eye sees only similarity.
What does it tell us about these two scribes themselves?
That they likely trained together or in the same tradition. Their styles are so similar that without computational analysis, most people would assume one hand. But the second scribe was less consistent, more variable. Maybe he was younger, or less experienced, or simply had a different temperament.
Could this method work on other scrolls?
That's the real promise. If you can identify individual scribes across multiple manuscripts, you start to see patterns—who worked when, how scribal practices changed, whether certain people collaborated. You move from studying texts to studying the people who made them.
Does knowing two people wrote Isaiah change what the text means?
Not the meaning of the words themselves. But it changes our understanding of how these sacred texts were produced—whether as a single act of devotion or as collaborative labor, whether the work was rushed or careful. It's the difference between knowing a book was written and knowing something about the person holding the pen.