Wholes, as opposed to collections of parts, are united.
What is it to be a whole? For a thing to be a whole means that it is united: a single thing. Although it may be composed of other things, and it may in turn compose other things, there is some integral quality to it.
For a thing to be an integrated whole also seems to imply that there is something else with which it can be differentiated. Wholeness is the result of a boundary: it makes everything inside the boundary the same thing, and everything outside a different thing. Everything is not like this, in that it does not have an inside and an outside (these are qualities only of something ).
Wholes do not generally overlap one another, although things can be at least partially coextensive with other things (i.e. they can occupy the same space). For example, this coextensiveness is allowed when the material is the same: the top 2/3 of my body is partially coextensive with the bottom 2/3 of my body. In this case, it is not the things that overlap, but the references to them. Overlap is generally not allowed when the material constituting the things is not the same. In fact, it is not even clear what it would mean for different material to occupy the same space, unless one of the things is tangible and the other is some sort of an ethereal, ghost-like thing.[9]
Wholes are said to be greater than the sum of their parts, which is a bit of a puzzling notion. There is at least one way in which it is true, and one in which it is false. On one hand, a whole is not greater than the sum of its parts if we understand identity materially: the material that composes a whole is exactly the material that composes its parts. For example, the material that constitutes the wheels, body, and the rest of a car is equivalent to the material that constitutes the entire car. On the other hand, a whole is greater than the sum of its parts if we consider properties such as the relations between parts to be properties of the whole, and not properties of the parts themselves (e.g. the spatial arrangement of parts). For example, the wheels, body, and the rest of a car are not sufficient to carry you about if they are lying in a heap.
A further claim about what makes wholes more than just the sum of their parts has to do with emergent properties . Emergent properties are said to emerge only when considering the whole, and are not properties of the parts of that whole. An example of this claim is that simple neuronal elements connected in varying ways leads to a brain: a whole which has properties that are not properties of the individual elements. Whether or not these properties could have been predicted based on knowledge of the smaller parts is the subject of some debate. In either case, the behavior of the whole can certainly be quite difficult to predict based on knowledge of the individual elements.[10]
In summary, there is something cohesive about things. Physical things tend to be cohesive in that three-dimensional objects often maintain their (approximate) shape. By contrast, the labels which we apply to these things are even more cohesive: objects are changing all the time, but their names are not.[11] Although using the same word for slightly different objects leads to an economy of expression, it is prudent to ensure that this categorical understanding does not reduce our relationship with reality to one which is exclusively categorical.
[9] We assume that if things do not occupy the same space, then they do not occupy the same space at a small physical scale, either (which rules out mixtures of things).
[10] For a popular example, visit the boids link at http://www.cognitivesettheory.com/links
[11] Thank goodness. I have a problem even with names that are not changing all of the time.