Where does float collide?

In the critical path method, the total float of an activity is the amount of time it can be delayed without pushing back the finish of the whole project. Kelley and Walker's 1959 Critical-path planning and scheduling set out the method and computed, for each activity in a project network, the gap between how late it may start and how early it may start, which is this spare time. An activity on the critical path has zero float, while the others carry a cushion a scheduler can spend. The referent is a slack interval in a project schedule, counted in working days, and it has nothing to do with glass floated on molten tin or with the way a computer stores a number.[1]

Unit

Spare time on one activity, in working days

Recovery means

Not applicable

Measured by

Computed from the project network as the activity's latest allowable start minus its earliest possible start

Float glass is a flat, even sheet of glass made by delivering molten glass onto a bath of molten tin, on which it spreads out and floats to a smooth, uniform thickness before it cools and hardens. Pilkington's 1969 The float glass process described the method he and his colleagues developed at Pilkington Brothers, in which the glass rides on the denser liquid tin so that both faces come out flat and parallel without any grinding or polishing. The word here names both the process and the product, a pane whose flatness comes from having floated. The referent is a manufactured material and the method that makes it, not spare time in a schedule or a format for storing numbers.[2]

Unit

A continuous ribbon or cut sheet of tin-floated flat glass, by thickness and area

Recovery means

Not applicable

Measured by

Assessed by thickness uniformity, optical flatness, and surface quality of the finished sheet

Floating-point is a way of representing real numbers in a computer as a fixed count of significant digits scaled by an exponent, so that the radix point can move to cover very large and very small magnitudes with the same format. Goldberg's 1991 What every computer scientist should know about floating-point arithmetic is the standard reference on how such numbers are stored, rounded, and made to behave predictably under the IEEE 754 standard. Because only finitely many values can be held, most real numbers are approximated, and the format trades exactness for range. The referent is a numeric data format inside a machine, described by its precision and its rounding, and it is unrelated to slack in a project or to glass floated on molten tin.[3]

Unit

One machine number: a sign, a significand, and an exponent at a given precision

Recovery means

Not applicable

Measured by

Characterised by precision in bits, the base, and the rounding behaviour defined by IEEE 754