More Options

The Harper’s Mansion Ghost Study

Paul McCaffrey

Skeptical Briefs Volume 24.3, Fall 2014

There are countless ghost hunters around the world, most of them doing what science writer Sharon Hill would dub “sciencey things.” Here we present a case study in bringing more scientific methodology to the pursuit. –Ed.

The following contains excerpts from a study completed in 2013 for Ghost Hunters Wollongong (GHW). We are an Australia-based group of researchers and investigators interested in claims of the paranormal. This example was edited on request and does not reflect the full report and associated data provided to Benjamin Radford, the University of Wollongong, Harper’s Mansion staff, and our members. The full version is available for purchase via email request at .(JavaScript must be enabled to view this email address).


As early as 2010 we began exploring the ever-growing trend of popular paranormal television shows. With my background as a science educator, I soon found problems in the “science” claimed by these investigators. It seemed that there was very little time in which the scientific method was actually employed to corroborate their evidence. After many investigations we decided there had to be a better way of confirming the actual environmental conditions and any unexplained activity that were sporadically reported, such as temperature changes, unexplained light and dark anomalies, electronically captured noises, etc.

Harpers Mansion

The building of Harper’s Mansion began in 1834 by James Harper, a wealthy landowner in the town of Berrima, New South Wales. It has had many occupants since then and many claims of paranormal activities. These include an apparition of the ubiquitous “lady in white,” and psychics claim to have felt the presence of spirits at the location.

From our perspective, we had investigated the mansion twice previously and noted that we had experienced more so-called EVPs (electronic voice phenomena, or ghostly voices) here than at any other location. This seemed to make it a good candidate for investigation, as we were familiar with the environment.

Outside view of Harper’s MansionOutside view of Harper’s Mansion in Berrima, New South Wales, Australia.

In conducting this experiment we sought out a clearly testable hypothesis that could stand up to the requirements of scientific rigor. Though it was difficult with so many claims made by paranormal investigators, we wanted to narrow the field to easily recordable, scientifically relevant, and most commonly reported claims.

Therefore, our chief aim in conducting this experiment was: “We will show an increase in the number of recorded changes within the physical environment tested.”

Baseline Measurements

While most groups today conduct a baseline in situ on the day, or perhaps with some investigators walking around the location on a prior visit, this is grossly inadequate as an accurate record of the “usual” conditions that may exist on the night of an investigation. For some three months, each week prior to our start date, GHW set up our baseline in and outside the house to record the conditions during the time the experiment was to occur, and also over a forty-eight-hour period around the time of our investigation.

Equipment used included three data loggers to measure temperature, humidity, and atmospheric pressure and two motion-activated cameras. The cameras and loggers were placed in the same position as they would be for the actual experiment.

Experimental Controls

Cameras and loggers were placed both inside (upstairs and downstairs) and also outside the building in storage shed (data logger) some twenty meters from the main house. Each independently measured the changes occurring during a single forty-eight-hour period each week, for four weeks, totaling 192 hours of measurement.


Daily changes in temperature, atmospheric pressure, and humidity over any one forty-eight-hour period were recorded and graphed and showed no significant changes outside the normal daily fluctuations that occurred naturally and were compared to recorded changes in conditions by the Bureau of Meteorology. No visual changes were captured by our cameras.

The Experiment

The experiment itself consisted of running thirty pre-recorded questions that were looped to run twice over a twenty-seven minute period each night as our equipment recorded any changes within the physical environment.

Questions asked of any ghostly occupants were all different and presented a range of information we wanted to know. These would involve directly provoking personal information regarding the past occupants of the house (as this is claimed to be a way to illicit a response), general information about the present state of the house and town, and future predictions. Questions were divided into three sections and for each a set of ten questions were asked.


• Two mini DV cameras (infrared). One for filming the area leading from the bedroom to the hallway and our equipment, the other covering the investigators during the experiment.

Investigators’ view of bedroomInvestigators’ view of bedroom at Harper’s including central recorder, decibel meter, and speakers.

• Three data loggers measuring for atmospheric pressure, temperature, humidity as previously set up for the baseline.

• Two motion-activated cameras (infrared) as previously set for the baseline.

• Six digital voice recorders (Olympus VN-8100PC) and 1 central digital recorder (Olympus WS-811). This recorder was placed at the center of our experimental setup. Four were located north, south, east, and west of the central recorder. One was located downstairs, directly below the central recorder, and the last outside the building near the western side of the building closest to the road.

• One lux meter that recorded readings every second for the twenty-seven minute duration.

• One motion activated sensor that activated an alarm when movement was present (aka the “shadow detector” used by many paranormal groups).

• Two geophones located in both the hallway and in the bedroom.

• One decibel meter on a table located directly above the central recorder.

• One computer to provide a visual representation of the lux readings, while another computer delivered our audio of the questions to be relayed, in real time, from the bedroom (upstairs) to a speaker downstairs, and outside to another speaker.

Experimental Controls

• Our audio to ask questions via the pre-recorded computer file, simultaneously eliminated problems associated with timing when correlating data after the experiment. This also prevented any contamination/variance by the investigators asking the questions in person each time.

• Investigators were removed from the building for the second looping of the questions to further eliminate any human contamination.

• One camera running simultaneously to the other (from a separate room; see above) recorded our investigators to eliminate ourselves as the source of any visual or audio changes during the experiment.

• Six digital recorders acted as controls to decipher sounds picked up in the location that were not coming from the one designated “EVP” recorder located in the area where “ghosts” were asked to speak on each occasion. Once sounds were not pinpointed as coming from the central recorder, they were eliminated from the census of data as originating from another possible source.

• A decibel meter was placed above the central recorder to act independently as confirmation to any sounds picked out on our central recorder or for independent sounds to be recorded not from the central recorder (another claim made by paranormal investigators often that sounds are found on one device but not on another).

• A lux meter and various other visual controls as mentioned in the equipment section would act to verify any changes in light recorded or captured visually on cameras.

• The investigators were seated and stationary during the first part of the questions discounting sounds made by moving, as is often seen on TV shows purporting to show actual ghost voices.

• Investigators recorded observations on sheets of anything audible or seen during the questions to later compare to the recorded data.

• All doors were locked prior to, and following, investigators entering and leaving the building. Shades were drawn and rooms checked for stray light sources.

• Equipment was placed in the same spot for each session. Baseline equipment was placed in the same spot as the three months prior to starting the experiment.

• Sessions were run concurrently on the same nights of the week (Monday and Tuesday) at the same starting time (7:00 pm) Starting times are within a one-minute margin of error.

• All equipment was checked prior to each session for batteries, and files erased for each session. New tapes were used for each session in cameras requiring tapes. • Only investigators were present during controlled conditions.

• A computer-generated silence of ten seconds was inserted between each question (and for 30 seconds at the end of each section of questions) to further eliminate contamination of the audio.


Over four months from May 2013 to August 2013, we conducted a total of twenty-seven sessions and approximately twelve hours of recording. Our audio recorded only four anomalies, which were found to be “drop-outs” on the audio, most likely due to the computer’s screen saver kicking on, and no increase in responses was recorded. No audible responses were detected and recorded while the investigators were present.

Visual evidence of momentary light fluctuations was restricted to one evening, June 25, occurring concurrently with recorded vibrations picked up by our geophones. At this time the town was experiencing the heaviest rainfall (and subsequent wind squalls) of the month. Investigators noted we could “feel” the natural movement within the building and hear door frames rattling.

chart showing light levelsLight levels stayed constant except for when investigators left at 7:12 PM.

There were no variations of the light levels from the baseline recorded during our sessions except when the investigators left the building walking past the meter. An example is shown above in which investigators leaving can be seen at 7:12 pm.

The graph below is an example for the decibel meter. It illustrates three sets of ten questions, represented by a peak and trough, during the experiment. Each trough represents the baseline for the room on the evening recorded and is ten seconds long. No unexplained disturbances were indicated over the period of the experiment.

chart showing sound levelsNo unexplained disturbances in decibel readings were indicated throughout the experiment.

Atmospheric pressure, temperature, and humidity also remained constant for the test periods as shown before in the baseline. No “bunching” of data occurred that would indicate sudden and/or sustained drops or rises in these measurements over a short period of time, as is often claimed by paranormal groups.

Including these eight changes to the physical environment existing within the controlled period of the experiment, changes amounted to 0.5 percent of the total data collected during each of the 1,620 questions asked where a possible response could be given under controlled conditions. These changes could easily be interpreted as a margin of statistical error on our part when conducting a wide-ranging experiment over such a length of time.

For the true believers, here is just a tiny portion of our debunks, which were easily explained away through our recording coverage as occurring in other locations, rather than in the room itself.

C. Rec
S1 P1 Q7
Noise. Was a car.
Checked West, Outside, and East.
C. Rec
S1 P1 Q8
Laugh. Heard on West. Stomach noise.
C. Rec
S1 P2 Q8
C. Rec
S1 P3 Q2
No? Heard on West recorder as well.
C. Rec
S1 P1 30s
Definitely stomach noises. Heard on West.
C. Rec
S1 P2 Q4
Siren? Car/ Heard outside.
C. Rec
S1 P3 Q7
Paul breathing. Heard on West.
C. Rec
S1 P3 30s
Talking? Heard on outside.
C. Rec
S2 P1 Q2
BANG. Heard on outside.
C. Rec
S2 P1 Q7
That same wood cracking noise we heard all night. Happens several times. Heard on several recorders.
C. Rec
S2 P2 Q5
Couple of footsteps. Just settling noises. Heard on other recs.
C. Rec
S2 P2 Q9
Female response? Blue? Heard on West as well.
C. Rec
S1 P2 Q3
Noise. Voice/noise after Q. Debunked, heard on other recs.


While no evidence was found to support any kind of paranormal activity at Harper’s Mansion, we do not intend our work to conclude that this would rule out any evidence at other locations. Rather, our work was designed as a field study to be applied at different locations as a more accurate way of scientifically measuring the existence of any such phenomena if they exist. We feel that as long as investigators will continue to believe in the paranormal and investigate—and they will—then why not try to solve the mystery once and for all.

Imagine you have just moved to a new home and you are settled in bed. Perhaps peering into the dark nooks of the house can any of us imagine that we would hear or see unfamiliar sights and sounds and find them mysterious? On ghost hunting vigils we are acutely aware of the signals our senses are receiving, in fact that is the point, straining to hear and see (usually in the dark), in an unfamiliar location. It is no wonder we find the unusual and unexplainable.

I think I can speak for the group when I say we all remain open to any ideas, but if it is real answers you desire, then better methods need to be applied than the ones we see offered to the public in today’s popular beliefs on ghost hunting and the paranormal. Some simple science can go a long way in solving the quite real fears people often have in these cases. Good luck in your future investigations.


Many, many thanks to Meagan Archibald (lead investigator GHW), John Keith Walker-Smith (treasurer and committee member of the Harper’s Mansion Management Committee, National Trust of Australia NSW), Danny Grammel (investigator), Frank Oliveri (investigator), Ben Radford (author, columnist, investigator and research fellow with the Committee for Skeptical Inquiry and deputy editor of Skeptical Inquirer science magazine), Assoc. Prof. Rodney Vickers (head of physics, associate dean—education & faculty of engineering and information sciences, University of Wollongong), and David Borger (physics teacher).

Paul McCaffrey

Paul McCaffrey is founder and coordinator of Ghost Hunters Wollongong (GHW).