top of page

The Spectral Network

Since the Spectral Network graphs relationships that possibly never occurred, it allows for a different study that the other networks on this site.  First, it explains how the people involved in the Salem Witch Trials experienced the trials.  Second, it demonstrates the strategy used in accusations by depicting how accusations succeeded.  Third, it shows how confessors behaved during the trials.  These three pieces provide ample data to prove strategic action during the accusations. 

When people heard the claims from the accusers that Goody Cloyce and Goody Procter afflicted them, it crafted a narrative of the activities of the witches.  The descriptions of witches' sacraments and baptisms created the negative image of the suspects through association.  The Usual Suspects Group on the right of Figure 7 shows a tight knit cluster.  The center of that group includes most of the Salem Village suspects and the Andover confessors (Sarah Good, Rebecca Nurse, George Burroughs, John & Elizabeth Procter, Giles & Martha Corey, Mary Lacey Jr., and Richard Carrier, although Martha Emerson and Mary Bradbury are also in the circle).  This center connects strongly to the green group, specifically three nodes.  The green nodes Mary Toothaker, Martha Carrier, and Ann Foster dominate the bottom group, but reach to the blue group.  What is happening here?

Salem accusations happened before Andover accusations, so the Salem suspects must be considered separately.  Good, Burroughs, Nurse, the Procters, and the Coreys appeared a month before Martha Carrier and a couple months before Toothaker and Foster.  Richard Carrier and Lacey Jr. confessed much later as well.  However, all their specters connect strongly in the network.  Just after the first accusation against Martha Carrier, and a month before the Andover accusations started, Bridget Bishop hanged for witchcraft on June 2, 1692.  The colony accepted the existence of witches, and acted against this perceived threat.  When considering the language of the witches' sacraments, it makes sense for suspects like Good, Nurse, and Burroughs to appear in the tight knit network.  Good served as a deacon, Nurse sat at the upper end of the table, and Burroughs was painted as the ringleader.  These accusation established the Salem witches as part of a hierarchy.  In order for later accusation to successfully pass, they needed to reference the suspects established as witches to sound realistic.

The Spectral Network and Edge Weight

The connections between town suspects only starts the strategy performed by the accusers. In order to make a believable accusation, suspects needed to appear with others.  Accusing Rebecca Nurse might pose certain difficulties, but when testimony against Nurse includes five references to seeing her and the Devil or four references to her and George Burroughs, it crafted a more suspicious narrative to challenge doubts about Nurse’s guilt.  Four or five references might not seem like a lot in the context of the witch trials.  However, four or five references in comparison to most of the relationships mentioned.  The graph below shows the edge weight for the spectral network.  Unlike the Master and Newton Networks, the edge weights here are based on the intentional act of speaking a pair of names rather than happenstance of being referenced.  This makes the edge weight most relevant in this graph since it was controlled by the accusers.  The 18 edges marked on the graph is slightly skewed since the John and Elizabeth Procter were married they were more often accused together as well.  However, a vast majority of the spectral pairs appeared together with limited frequency.

The accusations themselves had intention behind them to target specific people, but so did the claims of specters visiting together.  Nearly 2/3s of the pairs sighted appear one time.  The pairs that occur four to eight times are much fewer in number which suggests much more interest in naming these suspects.  A number of these higher number edge weights come from the Andover confessors.  Richard Carrier, Mary Lacey Jr., and Mary Lacey Sr. These instances are confessions where they admit to appearing with other witches, mostly each other as well as naming Martha Carrier.  Through these confessions, they manage to name Martha Carrier and the Devil as appearing 11 times, the second highest edge weight after the Procters.  The narrative created from this gave Martha Carrier the name the Queen of Hell.

The other group of specters that appear frequently is the Salem "usual suspects" which would include Nurse, Good, Osborne, the Procters, and Coreys.  The confessors all associated themselves with the Devil, and this group of suspects shares that association, but Rebecca Nurse's appeared with the Devil the most of the Salem accused.  Her specter was seen five times with the Devil while the next, Sarah Good, was four times.  It is interesting to note that of the Salem accused, Cloyce and Nurse were the most frequent pair followed by Osborne and Good, then Nurse and Martha Corey.  Rebecca Nurse's specter appeared so frequently since she needed to as a way to convince people of her guilt.  If any accusation were to fail, it would have been Nurse's.  Few people thought anyone could accuse her of witchcraft and so many people defended her that testimony of her specter causing mayhem helped sow doubts about Nurse's innocence.  The social network analysis demonstrates a clear effort to tie her to other suspects by placing her in the group of usual suspects and her frequent appearance.

The above image shows the node representing the Devil/Black Man.  As the network statistics show, he is the most connected and most influential in this network.  Not only did the accusers give power to an imaginary figure, but his role was strategic.  Just in this image, there is a clear wheel and spoke pattern, more evident looking at the light blue group in Figure 9.  A fair number of the Devil's edges in this group connect only to the Devil or others around him.  Several of these are confessors like the slave Candy and the Barker family, but one stands out.  On the above image, the node extending at about 2 o'clock from the Devil by itself is Susannah Martin.  Martin hanged for witchcraft after a trial, yet her specter never appeared with other suspects unlike all the other convicted suspects who appeared in testimony more frequently.  Why Martin, who in the Master Network has 73 degrees, saw such drastic difference indicates the necessary testimony for an accusation.  Someone like Rebecca Nurse must be tied to other specters to create a convincing narrative, but Martin faced witchcraft accusations before, so a reasonable doubt of her innocence persisted.  The only narrative to create around her was her tie to the Devil. 

The Spectral Network and Eigenvector Centrality

The Eigenvector scores of the network show how intentional some accusations were through the trials.  The top three are the Devil/Black Man followed by George Burroughs followed by Rebecca Nurse.  Excusing the Devil for his non-existence, it leaves Burroughs and Nurse as the two significant ones to look at for comparison.  Testimony against Burroughs referred to him as the King of Hell and appeared at the Devil's sacraments sitting at the "upper end" of the table.  The language about these two suspects provides insight into the obvious fact.  Burroughs and Nurse were desirable targets to convict.  There is an entire history of the Putnam family's conflicts with Rev. George Burroughs, and there certainly was an effort by the Putnams to accuse Rebecca Nurse.  However, the "why accuse them" question has been covered by several historians, so how were they accused?  A Puritan minister and a respectable church member were challenging targets to accuse.  Rebecca Nurse nearly escaped execution at trial.  The language used by the accusers turned people against these suspects.  The fears at play during the period made the narratives of the accusations take shape against people previously untouchable with such claims.  However, placing George Burroughs as the Devil's second in command and Rebecca Nurse as an important figure reshaped popular imagination in their neighbors' minds.

Given George Burroughs' previous challenges in Salem Village, there only needed to be enough doubt which the accusation probably provided on its own merit.  The continued stories about Burroughs continued to spill out of lingering anger at him.  The case of Rebecca Nurse is what shows how the court flipped opinions based on accusations.  At her first examination in March, Judge Hathorne remarked about Nurse, "It is very awful to all to see these agonies [the afflictions] & you an old Professor thus charged with contracting with the Devil."  Yet just a few months later in June, the jury found Nurse not guilty and "the honored Court was pleased to object against it."  This is how deep the accusations challenged the status quo of Massachusetts.  The unfortunate sight to see a church member accused of witchcraft turned into a sight of horror at her acquittal.  Nurse's case is unique since she was the only suspect in 1692 found not guilty, although the jury reversed their decision.  However, her case is also the only one with a similar regret expressed by the judges in the beginning.  The narrative created through the Spectral Network created that opinion.

The Spectral Network and Accusations

The image above shows Groups 3 and 4.  These two groups demonstrate important aspects of the accusations made during the witch trials.  Group 3 includes most of the Wardwell family as well as other suspects like Jane Lilly, Stephen Johnson, and William Barker Jr., all of whom were accused in September in Andover. This group is selected since it highlights one of the key themes of confessions better than other groups.  Notice the pattern of a rectangle with an X inside it.  It appears in groups 3 and 4 as well as others. Even with intergroup edges shown (see Figure 7) the box pattern is mostly isolated.  This is showing how accusations made through confessions occurred.  On September 1, 1692, the Wardwells, Barker, and Johnson confessed to witchcraft.  Much like other confessors, the people they named as fellow witches were those already accused, for the most part those in the room.  Confessions contained accusations, they did not spread them.  When people confessed, they broadly only confirmed suspects already arrested.  This means confessions did not cause the trials to continue spiraling out of control.  More on this can be found in the Accusations analysis.

The Escaped Group is included as a testament to this method of analysis.  The group includes Daniel Andrews, George Jacobs Jr., Philip English, Mary English, and Elizabeth Colson.  These suspects successfully escaped.  Others fled as well, but these were the recognizable names of those who ran away.  Only the list of spectral pairs were put into NodeXL to graph and group the names.  No variables such as escaped or family names were given.  Aside from showing how the program connects this information, it reveals something about the accusations by the accusing girls.  NodeXL does not randomly assign these clusters.  The accusations needed to include these connections for the program to find it.  This means accusations included part of this strategy to name suspect specifically in the context of others.  Specters seen together were not coincidental.

bottom of page