The width of the spectral line ΔE presumably comes from a series of measurements, each having some inherent uncertainty, plus some deviation due to a lack of experimental precision. Therefore we can expect that a measured ΔE will usually be larger than the minimum allowed by the Heisenberg uncertainty principle. A larger value of ΔE leads to a smaller Δt, because of the inverse nature of the relationship, and the true value of Δt will be larger.