The H-Index Was Not Designed for History
Jorge Hirsch introduced the H-Index in 2005 — fifty years after Einstein's death. The metric was designed to evaluate living researchers in active fields with well-indexed publication records. Einstein's most important papers were published between 1905 and 1935, long before systematic citation indexing existed.
This creates a fundamental problem: the H-Index requires that citations be formally recorded in a machine-readable database. The thousands of papers that built on special relativity, general relativity, and quantum mechanics in the mid-20th century were often not formally linked back to Einstein's original papers in any database. They were treated as common knowledge — a phenomenon bibliometricians call obliteration by incorporation: when a discovery becomes so foundational that it is cited implicitly rather than explicitly.
Why Different Databases Give Different Numbers
The variation in Einstein's H-Index across databases illustrates a principle that applies to every researcher — not just historical ones.
| Database | Coverage | Einstein H-Index | Why |
|---|---|---|---|
| Web of Science | Journals post-1970s, strict | 4–5 | Pre-1960 papers mostly unlinked |
| INSPIRE-HEP | Physics literature, deeper historical coverage | ~18 | Better retroactive linking for physics |
| Google Scholar | Broadest: books, preprints, grey literature | 100+ | Captures textbook and indirect citations |
Sources: Shah & Jawaid (2023, PMC) for Web of Science value; Hirsch (2005) for INSPIRE context; ResearchGate academic discussion for Google Scholar estimate. Values are approximate and change as databases update their coverage.
This is not unique to Einstein. Studies consistently find that the same researcher can have a Google Scholar H-Index 30–50% higher than their Scopus H-Index, simply because Google Scholar indexes conference papers, preprints, and books that Scopus excludes. For humanities researchers, the gap is even larger because books dominate their citation footprint.
Stephen Hawking and Richard Feynman
Hawking and Feynman are more accessible to bibliometric analysis because their careers extended into the era of systematic citation indexing. Hirsch himself cited Hawking's H-Index as approximately 62 in his original 2005 paper (using Web of Science), placing him among the most impactful physicists of the 20th century. Google Scholar estimates are considerably higher, for the same coverage reasons described above.
Feynman presents a different problem. His most famous contributions — Feynman diagrams, the path integral formulation of quantum mechanics — are so deeply embedded in modern physics that they are rarely cited as the original 1949 paper. Every physics paper that uses Feynman diagrams implicitly builds on his work, but almost none cite the original paper explicitly. This systematically compresses his measured H-Index relative to his actual scientific influence.
What This Means for Your Own H-Index
The Einstein example is extreme, but the underlying dynamics affect every researcher:
Always specify the database. An H-Index without a source is ambiguous. "My H-Index is 18" means something different if it comes from Google Scholar versus Scopus versus Web of Science. For grant applications and job interviews, most committees now expect the source to be stated.
Field matters more than the number. A physicist with H=18 and a humanities scholar with H=18 are not equally productive by any reasonable measure — the citation cultures are fundamentally different. The H-Index only meaningfully compares researchers within the same discipline and career era.
The metric rewards breadth over depth. Einstein and Feynman both produced relatively few papers with enormous influence. The H-Index does not capture this well. A researcher who publishes 20 moderately-cited papers will typically outscore one who publishes 5 highly cited papers, even if the latter's work is more significant.
Use our H-Index Calculator with live OpenAlex data, then compare your score against field-specific percentile benchmarks.
→ H-Index Calculator — live OpenAlex data → What H-Index puts you in the top 10%? 2026 Field Benchmarks → What is a good H-Index by career stage?What was Einstein's H-Index?
Einstein's H-Index varies significantly by database: approximately 4–5 in Web of Science (limited historical coverage), ~18 in INSPIRE-HEP (physics database with better retroactive linking), and over 100 in Google Scholar (broadest coverage including books and indirect citations). The variation reflects database coverage, not measurement error.
Why is Einstein's H-Index so low in some databases?
Web of Science has limited coverage of publications before the 1970s. Einstein's most important papers were published 1905–1935, before systematic citation indexing. Additionally, his work has undergone "obliteration by incorporation" — it became so foundational that it is cited implicitly rather than explicitly.
What was Stephen Hawking's H-Index?
Hirsch (2005) reported Hawking's H-Index as approximately 62 using Web of Science, placing him among the most cited physicists of the 20th century. Google Scholar estimates are considerably higher due to broader coverage.
Can you compare scientists from different eras using the H-Index?
Not reliably. The H-Index systematically undervalues historical scientists due to incomplete database coverage of pre-1970 literature, obliteration by incorporation for foundational work, and differences in citation culture between eras. It works best for comparing researchers within the same field and career period.
Which database gives the most accurate H-Index?
No single database is universally "most accurate" — each has different coverage. Google Scholar is broadest but includes grey literature. Scopus and Web of Science are more conservative but better curated. OpenAlex is the only fully open option. For most purposes, state which database you used alongside the H-Index value.
// Sources & References
- Hirsch, J. E. (2005). An index to quantify an individual's scientific research output. PNAS, 102(46), 16569–16572. [Hawking H=62 from WoS] doi:10.1073/pnas.0507655102
- Shah, F. A., & Jawaid, S. A. (2023). The h-Index: An Indicator of Research and Publication Output. Pakistan Journal of Medical Sciences, 39(2). [Einstein H=4–5 from WoS] doi:10.12669/pjms.39.2.7398
- Bornmann, L., & Daniel, H. D. (2007). What do we know about the h index? JASIST, 58(9), 1381–1385. doi:10.1002/asi.20609
- OpenAlex (2026). Open bibliometric database. openalex.org