The unfortunate part of the natural process of refinement of time scales is the appearance of circularity if people do not look at the source of the data carefully enough. Most commonly, this is characterised by oversimplified statements like:

"The fossils date the rock, and the rock dates the fossils."
Even some geologists have stated this misconception (in slightly different words) in seemingly authoritative works (e.g., Rastall, 1956), so it is persistent, even if it is categorically wrong (refer to Harper (1980), p.246-247 for a thorough debunking, although it is a rather technical explanation).

When a geologist collects a rock sample for radiometric age dating, or collects a fossil, there are independent constraints on the relative and numerical age of the resulting data. Stratigraphic position is an obvious one, but there are many others. There is no way for a geologist to choose what numerical value a radiometric date will yield, or what position a fossil will be found at in a stratigraphic section. Every piece of data collected like this is an independent check of what has been previously studied. The data are determined by the rocks, not by preconceived notions about what will be found. Every time a rock is picked up it is a test of the predictions made by the current understanding of the geological time scale. The time scale is refined to reflect the relatively few and progressively smaller inconsistencies that are found. This is not circularity, it is the normal scientific process of refining one's understanding with new data. It happens in all sciences.

If an inconsistent data point is found, geologists ask the question: "Is this date wrong, or is it saying the current geological time scale is wrong?" In general, the former is more likely, because there is such a vast amount of data behind the current understanding of the time scale, and because every rock is not expected to preserve an isotopic system for millions of years. However, this statistical likelihood is not assumed, it is tested, usually by using other methods (e.g., other radiometric dating methods or other types of fossils), by re-examining the inconsistent data in more detail, recollecting better quality samples, or running them in the lab again. Geologists search for an explanation of the inconsistency, and will not arbitrarily decide that, "because it conflicts, the data must be wrong."
If it is a small but significant inconsistency, it could indicate that the geological time scale requires a small revision. This happens regularly. The continued revision of the time scale as a result of new data demonstrates that geologists are willing to question it and change it. The geological time scale is far from dogma.

If the new data have a large inconsistency (by "large" I mean orders of magnitude), it is far more likely to be a problem with the new data, but geologists are not satisfied until a specific geological explanation is found and tested. An inconsistency often means something geologically interesting is happening, and there is always a tiny possibility that it could be the tip of a revolution in understanding about geological history. Admittedly, this latter possibility is VERY unlikely. There is almost zero chance that the broad understanding of geological history (e.g., that the Earth is billions of years old) will change. The amount of data supporting that interpretation is immense, is derived from many fields and methods (not only radiometric dating), and a discovery would have to be found that invalidated practically all previous data in order for the interpretation to change greatly. So far, I know of no valid theory that explains how this could occur, let alone evidence in support of such a theory, although there have been highly fallacious attempts (e.g., the classic "moon dust", "decay of the Earth's magnetic field" and "salt in the oceans" claims).

Carbon-dating of wood under lava that was erupted from Rangitoto (an island volcanoe near Auckland, New Zealand) indicates that the eruption was around 200 years ago. The name is said to mean red sky', suggesting that the Maoris, who have been there for 1,000 years at the most, witnessed this event.........

yet...potassium-argon dating of the lava has given ages of up to half a million years!

There is no way simply to look at a fossil and say how old it is unless you know the age of rocks it comes from. Sometimes igneous rocks, rocks we can date chemically, intrude into sedimentary rocks, and in such a fashion some hard-core “absolute” dates—expressed in terms of millions of years—are available for all subdivisions of geologic time. The earth is 4.55 billion years old (give or take a few million; the date comes from moon samples, meteorites and graphic extrapolation from rocks dated directly earth). The oldest rocks dated on earth are about 4 billion years old. We know that the Devonian Period [that includes the time period that Eldredge studied] began about 408 million years ago and ended roughly 360 million years ago. The middle Devonian came in about 380 million years ago.

But none of that helps in a cow pasture in upstate New York. Long before radioactivity was known to physicists, paleontologists had another way to tell time. Fossils occur in the same vertical sequence thoughout the geologic column. The same, or closely similar fossils frequently occur in many far-flung localities; some are even found worldwide. This repetitive pattern of occurrence allows geological minded paleontologists to correlate: rocks are mapped, and frequently certain distinctive horizons, such as volcanic ashfalls, can be traced over great distances. But rocks in isolated quarries can be matched up according to the nature of the fossils they contain. And this poses something of a problem: if we date the rocks by their fossils, how can we then turn around and talk about patterns of evolutionary change through time in the fossil record? We need an independent time frame to know that a trilobite in Ohio is roughly the same age as one in New York before we can talk about geographic variation; otherwise, their differences might as well be ascribed to the sort of process of gradual change that Darwin thought was inevitable with the simple passage of time. The distinction between the two—gradual temporal transformation of an entire species versus geographical differentiation within a species—is crucial, and indeed underlies the very notion of punctuated equilibria.