There's several official databases for hurricanes. The one I'm most familiar with is called HURDAT, which stretches back to 1851 for the Atlantic. Source: http://www.nhc.noaa.gov/data/.

The data isn't technically "complete" until the satellite era, but there are reports going back to 1851 where ship data and island reports can estimate cyclogenesis and based on reports from those ships and Caribbean locations we can estimate intensity. To a degree it's predictable, because an eye passing over is easy to note. When it isn't, you can extrapolate based on recorded minimum pressure and wind speed and direction what the eye would have been. Landfalls can be tracked back even longer when they hit populated areas, but there are far fewer ship reports.

This does not mean HURDAT is complete. It's not. It's completely possible for especially weak storms to have never been detected pre-satellite era. But it's unlikely a storm of this intensity wasn't recorded in HURDAT because the factors that allow for such intensities as Irma are also found closest to the Caribbean Islands. That does not rule it out, but there is no evidence for a hurricane stronger than Irma in the part of the Atlantic it is in.