The abstract of the study described here is part of a larger study on Gospel Parallels. The purpose of this study is to establish an empirical model for talking about the extent to which parallel passages in the Gospels are more or less similar to each other. In other words, everyone knows that the four Gospels differ from each other when they discuss common events. Some estimates are that there are more than 40,000 such differences, the most commonly referred to difference being that the Gospel of John has the tempest in the Temple at the beginning of Jesus’ ministry while the synoptic gospels have it at the end.

The main methodology is to take various scenes that appear in all four gospels and to break down the stories into their complete component parts, and then determine whether or not the information contained in each of the Gospels matches [1]. For example, all four Gospels describe the empty tomb. While they all agree that the visitors came to the tomb on Sunday (“the first day”), they don’t agree when. Mark (16:2) claims “the sun is risen” while Matthew (28:1) says that it “began to dawn” and Luke (24:1) seems to agree, describing it as “at early dawn,” but John (20:1) says it is “still dark.” Using our current model, we take all of these elements (e.g., day, time, etc.) and create a matrix that compares and contrasts the information to develop a mathematical description of the extent of the agreement.

We can do this at several levels. We can look at the overall agreement for a given scene (e.g., the empty tomb, the crucifixion, the anointing) or for a given Gospel writer (Mark vs. Matthew vs. Luke vs. John). We can examine not only agreement, but also disagreement, which is not necessarily the reciprocal of agreement due to the fact that not all Gospel writers comment on all issues (e.g., Neither John nor Luke mention any message from Jesus to the women). The possibilities are endless, and rely upon the interests of the investigator.

For our purposes here, we are interested in examining the claim that the anointing stories in the various Gospels are referring to several different episodes because of the differences between the scenes. In other words, some scholars have claimed that there was more than one anointing, and one of the bases for this claim is the differences between the scenes. For example, Luke sets the scene in Nain at the home of Simon the Pharisee while Mark and Matthew set the scene in Bethany at the home of Simon the Leper. John agrees that it was in Bethany, but puts it at the house of Lazarus. So how many anointings were there. One? Two? Three?

There is no way to make a judgment about this issue unless one establishes a base rate of agreement for Gospel stories. Remembering that the Gospels were written down many decades after the fact, we should expect a reasonable amount of disagreement between them just from what we know about the reliability of oral testimony. But in addition, we know that the Gospels were composed for different doctrinaire purposes, and this can contribute to the lack of agreement. Hence, it’s important to understand to what extent the Gospels differ in reporting the same story. Once we have this answer, we can ask the next logical question, which is - “Do the anointing accounts differ more or less than the general degree of agreement among Gospel stories,  and if so, how?”  If the anointing scenes differ significantly more than the normal degree of agreement, then perhaps the scholars’ theory that there is more than one story is correct, and the low degree of agreement is a symptom of this fact.

The anointing stories were broken down into their component parts. There were 21 different elements, including when, where, the name of the homeowner, the homeowner’s occupation or status, the name of the woman, the substance, etc. Then each Gospel account was compared to every other Gospel account, in a matched comparison method (Mark vs. Matthew, Mark vs. Luke, Mark vs John, etc.). In this manner, a total of 126 comparisons were made (6 pairs for 21 elements). Looking at the matrix, it’s clear that on some issues (e.g., the sex of the person who did the anointing) there is complete agreement, but on other issues, there is substantial disagreement (e.g., the homeowner).

Before we look at the results for the anointing scenes, let’s discuss the general findings vis-à-vis the other Gospel stories. To date there are four Gospel stories that have been analyzed in this manner: the empty tomb, the crucifixion, Joseph of Arimathea, and the anointing scenes. By virtue of their complexity, these four stories have different component parts or elements, ranging from a high of 28 (crucifixion) to a low of 21 (anointing).

Across all four stories over all four gospels, the average total agreement (all four Gospels agreeing for 97 different elements) was a mere 12%. Against these averages, the anointing scenes agreed 19% of the time which was the highest degree of agreement among any of the stories (crucifixion and Joseph = 11%, empty tomb = 9%). In other words, when the anointing scene is compared across four parallel stories in the Gospels, the anointing story has more agreement than the other stories. These differences were significant at the .001 level (Chi square = 62.08, df = 9).

Some scholars may argue that the measure of total agreement (all four Gospels agree) is too harsh a measure since not all Gospel accounts address all elements. Thus a second measure was determined in which those elements not addressed in a given Gospel were eliminated from consideration, and only agreement/disagreement on those elements specified was measured, by using a paired comparison method. This second measure was labeled “paired agreement” to distinguish it from “total agreement.”

Adopting a less strict measure resulted in more elements to be compared, increasing the total pool of items from 388 (97 elements across 4 Gospels) to 582 paired comparisons (Mark vs. Matthew, Matthew vs. Luke, etc.). Using this method, the average agreement for paired comparisons was 26%, ranging from a high of 40% (anointing) to a low of 18% (Joseph) with intermediate values of 27% (crucifixion) and 19% (empty tomb). These differences were significant at the .001 level (Chi square = 62.63 df = 9).

As in the previous measure, the anointing scene has the most agreement of any of the four scenes, and both measures yielded significant differences. In both cases the anointing scene has nearly 50% or more agreement than the closest rival.

If agreement between Gospel accounts is a measure of historicity, which I believe it is, then the anointing scene has the highest degree of historical accuracy among the four Gospel stories examined to date. But we can use this method to provide another measure of historicity by using the number of “not applicable” comparisons as a measure of lack of historicity. In other words, we can assume that important elements associated with a scene are discussed by the different Gospel writers. They may disagree on the details on these items (e.g., the homeowner is Simon the Leper vs. Simon the Pharisee), but they all agree that it happened (in a house while the participants were at a table eating). But occasionally one finds elements that are raised in one Gospel that have no counterpart in the other Gospels. Here are examples of elements discussed in only one Gospel:

• Matthew claims that when they went to the tomb there was a “great earthquake (28:2).” No one else mentions the earthquake.
• Luke claims that on the way to be crucified Jesus talked with “the daughters of Jerusalem (23:28).” No one else mentions any talk along the way.
• John claims that Jesus is pierced in the side with a spear (19:34). No one else claims this.
• Mark claims that Pilate inquires whether or not Jesus is actually dead (15:44). No other Gospel account includes this information.

To some extent, we may be dealing with literary license here, rather than historical fact. One measure of this license is the percentage of items in the comparisons which have been rated not applicable, because there is no comparable passage in another Gospel. Using this measure, among the 582 paired comparison tests, there were 289 pairs that could not be scored, equal to 50%. In other words, half the material discussed in the four Gospel accounts is not found in the other Gospel accounts. Previously we discussed the fact that 26% of the paired comparison items indicated agreement. At this point we can note that the remaining 25% is composed of 21% disagreements and 3% neutral items (i.e., items for which there is neither complete agreement no complete disagreement).

Examining the “not applicable” items that compose 50% of the comparisons. the Gospel story with the highest number of these items was the Joseph scenes, with 74% of the elements not discussed in all four Gospels. Examples of these unique commentaries include the facts that Joseph is rich (only in Matthew 27:57), a secret disciple (only in John 19:38), assisted by Nicodemus who brings 100 pounds of aloes and myrrh (only in John 19:39), and the tomb belonged to him (only in Matthew 27:60). The story with the least amount of “not applicable” items was the anointing story, with a mere 26% of the items ranked accordingly. Intermediate values were obtained for the empty tomb (47%) and the crucifixion (47%). These differences were significant at the .001 level (Chi square = 46.64, df = 3)

To summarize, we developed three measures to examine the extent to which various stories agree across different Gospel accounts. Two of these measures involved the extent to which the elements in different Gospel accounts are in agreement, either on a total basis (all four Gospels agree) or a paired comparison basis. The third measure looked at the extent to which Gospel accounts have unique material not found in other Gospel accounts.

On all three measures, the anointing stories achieve the scores that indicate they have the least discrepancies. They achieve the highest scores on total agreement (19% vs. 11%,11%, and 9%) and the highest scores on paired agreement (40% vs. 26%, 19%, and 18%). On unique content, the anointing stories achieve the lowest score (26% vs. 47%, 47%, and 74%).

On this basis, we can dismiss the theory that the anointing story is a reflection of numerous anointing episodes because of the differences between the Gospel accounts. Our data show that the anointing stories have more agreement and less unique material than other Gospel stories. If differences between anointing stories suggest different anointing episodes, then by this same logic Jesus was crucified several times.

Having discovered significant differences between Gospel stories on agreement as well as uniqueness, we can ask what accounts for these differences? Historicity? Is it possible that accounts which have high agreement and low unique material reflect a true episode, while lower levels of agreement and higher incidents of uniqueness suggest non- historical material? Yes, this is certainly possible: even likely. For example, my own personal idea about Joseph of Arimathea is that he was a wholly invented character, a belief shared by John Shelby Spong (1994) and others. In our analysis, the stories about Joseph achieve the least agreement and have the most unique content, suggesting that these measures are a good indicator of historicity, at least in this case.

Is it possible that some factor other than historicity is responsible for these results? For example, if the Gospel stories are derivative of mythological stories with Babylonian or Egyptian origins, or re-writings of Old Testament tales, then is it possible that the anointing story holds up best because the stories upon which it is based were more well known that the other stories under study? Alternately, is it possible that the anointing tale has more agreement and less unique content because the original Markan story was followed more closely than the comparable stories, producing the data based on the strength of the original Markan account rather than a reflection of true history? Or are the differences found here merely an artifact of the methodology, and with further analysis of more stories, subject to disappearing?

Is it possible that unique content is a function of the time and place associated with the story telling? For example, if the Gospel of John has Alexandrian origins, are the unique details in the anointing tale a function of appealing to Alexandrian audiences .

The answer to all these questions is “Yes.” All these influences are possible; however, for our purposes in discussing Mary Magdalene, they go beyond the scope of our mandate. The original rationale for this study was to examine whether or not the discrepancies in the anointing stories in the four Gospels was a sufficient rationale for dismissing the theory of the “single” anointing story. Our study shows that, compared to other Gospel stories, the anointing story, in fact, has more agreement and less unique content. By itself, this finding is sufficient to dismiss the argument that there must have been two or more anointing stories because of the differences between the different accounts. This study doesn’t establish that the anointing story is necessarily true; however, to the extent that agreement and uniqueness of content are measures of historicity, it suggests that the anointing stories may be closer to an historical account. Further research needs to be done to link these measures to historicity directly.