The hair felt it. The antenna didn't.
This site's own buzz room is about what a bumblebee has to work out with her muscles once she's already gripping a flower. Real bumblebees read something else first, before they ever land: a floral electric field. Earth's fair-weather atmosphere carries a real, ever-present field of roughly 100 volts per meter near the ground; a flower, grounded through its own roots and stem, sits weakly negative against it. A foraging bee doesn't — friction against dust and air molecules during flight strips electrons from her body, so she typically arrives positively charged. That contrast alone puts a small electric field around every flower a bee approaches, and a 2013 Science paper (Clarke, Whitney, Sutton & Robert) found Bombus terrestris can discriminate that field's own pattern and structure, not just whether one is present, and learns a rewarding flower faster with it than with color and scent alone. The field is a live signal, too: the paper measured a petunia's own stem potential rise by about 25 millivolts the instant a bee lands, staying changed for just under two minutes — longer than one visit takes, short enough that the next bee can read "someone was just here" honestly, before the nectar itself could plausibly be topped up again.
None of that explains how a bee, with no evolved electroreceptor organ — nothing like a shark's ampullae of Lorenzini — actually feels a field with no mass and no smell. A follow-up in 2016 (Sutton, Clarke, Morley & Robert, PNAS) went looking with a laser Doppler vibrometer, comparing two body parts a bee already has: antennae and the fine mechanosensory hairs covering the rest of her. Both move under a bee-relevant field — Coulomb force pulls on any charged surface, hair or antenna alike — but only the hairs' motion showed a matching spike recorded straight from the nerve underneath; the antennae deflected with no electrophysiological response to show for it. The organ doing the sensing turns out to be the same fine hairs a bee already uses to feel touch and airflow, pressed into a second job neither the hair nor its nerve was necessarily built for first — an existing part reading a signal nobody had to evolve a new one to catch.
The honest gaps: the "just under two minutes" figure is one species (Bombus terrestris) on one plant (petunia) under lab conditions, not a number that has to hold for every bee on every flower in a real field; and "no evolved electroreceptor" describes what's absent, which is always the weaker kind of claim — it holds only as well as the searching that produced it. What both papers earn together, though, is a real mechanism end to end: a physical reason the field exists (grounding plus frictional charging), a behavioral proof it's used (bees discriminate and learn from it), and a specific structure identified as the sensor (hairs, not antennae) with electrophysiology to back that specific claim up. Three different kinds of evidence, not one.
Sources: Clarke, Whitney, Sutton & Robert, Detection and Learning of Floral Electric Fields by Bumblebees, Science 340(6128):66–69 (2013), doi:10.1126/science.1230883. Sutton, Clarke, Morley & Robert, Mechanosensory hairs in bumblebees (Bombus terrestris) detect weak electric fields, PNAS 113(26):7261–7265 (2016), doi:10.1073/pnas.1601624113.