Hot News


   Return to WHIRC Home Page


This page notes recent updates, news which might not yet be incorporated into the manuals or problems encountered during observing runs which will (hopefully) be repaired in the near future. We intend to update this frequently.

26 June 2014: The primary mirror was hand washed in February 2014, and the signal levels have increased by the order of 10%, to values generally larger than those ever seen in the past. These signals are up by 60 - 90% above those seen prior to the recoating of the WTTM mirrors.

In addition, Ron George, Charles Corson, and Bill Ball reterminated the DHE power cable and removed the "pigtail" from the system. This has resulted in consistently lower read noise and possibly some immunity to pickup from other power sources.

WTTM has been used frequently by observers with only occasional problems. We recommend that observers use WTTM if their programs will permit it (e.g., relatively small dithers, fields with sufficiently bright stars), since it does appear to improve the image quality and ameliorate the effects of telescope wind shake. Frequent use will also maintain the proficiency of the OAs in the use of this system. Jenny Power has written a web-based tool for locating WTTM guide stars, accessible from the main WHIRC webpage; this can be used to determine in advance whether suitable guide stars are available and their location in the WHIRC field of view.

10 December 2013: Charles has worked his magic again and now reports that the WTTM tip/tilt stage is working. Last night Anja Weyant (remote observer for the SN monitoring program) and Krissy used WTTM the entire night without any problems. So for now, we are back in the saddle again.

23 September 2013: Thanks to Charles Corson, Lori Allen, and Doug Williams, we have measured both standard star signals and flatfield lamp settings for WHIRC with the recoated WTTM mirrors. Basically, the signal levels have returned to those which were seen after the commissioning of WHIRC in 2007; slightly higher in some filters, slightly lower in others. The raw signal levels will vary around the array before flatfielding; the published numbers are the average of five measurements spaced over the detector.

During the T&E night, the image quality appeared quite sensitive to the focus, with noticeable astigmatism in defocused images. This has always been the case, but we will need to investigate further whether the sensitivity has increased with the reinstalled WTTM mirrors. In the meantime, observers are cautioned to keep a close eye on the image quality and adjust the focus as needed (the astigmatism provides a useful guide to the direction and magnitude of the required focus motion).

The WTTM tip/tilt mirror itself does not appear to be operating reliably in one axis, although the actuator holds the mirror in place. We will continue to look into this, but for now the active tip/tilt is not functioning.

The flatfields look very similar in illumination to those obtained prior to the WTTM mirror recoating, but we still suggest obtaining flats during each night of science observing, rather than relying on archival flats.

11 September 2013: The WTTM mirrors were recoated over the summer through the heroic efforts of Charles Corson, Gary Poczulp and Heidi Schweiker. Because of damage which occurred during shipping to the coating vendor, some of the mirrors originally in WTTM were stripped of their existing coating, for the others we were able to use spare mirrors from the original WTTM construction. The four WTTM mirrors, including the tip/tilt mirror, were electrocoated with a proprietary Laser Gold coating which can be cleaned if necessary. The WTTM flip mirror was recoated with a conventional evaporated overcoated gold layer because it is on a glass substrate and could not be coated with the Laser Gold. Reinstallation of the mirrors is underway. We anticipate that the throughput of WTTM should be increased with the new mirrors. We will attempt to measure standard star signal levels and flatfield lamp settings during the T&E on 19 September and update the manual with the new information.

Before Jenny Power left us for the LBT, she wrote an interactive tool for locating potential guide stars for WTTM operation which provides both a tabular and graphical output. Dither patterns can be overlaid to illustrate if a selected guide star is suitable for a given pattern. The tool can be found at http://www-kpno.kpno.noao.edu/cgi-bin/kpno-misc/wiyn/WTTMtool.pl which is also linked from the WHIRC homepage. Thanks to Doug Williams for installing this tool.

19 April 2013: During the recent HYDRA block, the WTTM module was taken downtown for inspection of the mirrors. As noted earlier, the tip/tilt mirror itself is in very poor condition, possibly as a result of the mask paint interacting with the antireflection coating. The other WTTM mirrors were dusty but in otherwise apparently good condition. All mirrors were cleaned with CO2 by Gary Poczulp. Removal and replacement of the tip/tilt mirror is in the plans for the summer.

After heroic efforts, Charles has managed to get WTTM running reliably. Issues with the X stage motor and controller, combined with an apparent resonance seemed to be the major factors. We are planning on-sky tests in the future.

Behzad has added a couple of new widgets to the MOP. The first of these is a confirmation widget which pops up when one closes the MOP, requesting a "YES/NO" reply. Previously, an inadvertent clicking of the "exit" button (which can happen particularly when observing remotely through VNC) put the MOP into an irreversible shutdown. A second widget will pop up with the message "Failed to acquire 'exposure name'. Skipping 'exposure name'", with an "OK" button when the PAN for some reason fails to deliver a complete exposure. While the exposure in question will be lost, this avoids the necessity to shut down and restart the PAN/MOP as had been necessary in the past.

2 October 2012: After observer reports suggesting that the WHIRC signals from standard stars have decreased, Lori was able to measure standard stars through all 13 filters on a photometric night last month. The results confirm that the signals have decreased significantly since WHIRC was installed on the telescope in 2008. The most likely culprit is deterioration of the WTTM mirrors, which were coated with a "protected silver" coating when they were new. While this was probably done to provide good performance in the R band with the original optical WTTM (avoiding the dip in Al reflectivity at 800 nm), it is likely that the years have taken their toll. Recoating the mirrors will not be a simple operation, but is something which we need to consider.

21 September 2012: Over the summer, pODI was installed at the WIYN port and the IAS was moved to the MOS port of the telescope, where it will share use with Hydra. The power supply cables for WHIRC were refabricated slightly longer to accommodate the extra distance from the supplies to the MOS port. Dark observations at the new location suggest that the read noise may be very slightly elevated compared to the value earlier in the year, possibly resulting from the longer power cable. However, this effect is very small and may not be significant compared to the normal fluctuations seen in the read noise level with time. Unfortunately, the significantly elevated noise seen with MiniMo powered up is still present, so joint WHIRC/MiniMo observations will require the latter to be powered off when WHIRC is in use as well as the roughly 30 minute warmup time once MiniMo is powered up.

Observations of M13 in the new location show that the existing WCS solution is adequate, with a possible 0.2 degree rotation offset of the field. For now, we plan to make no changes to the WCS constants.

10 February 2012: The new CN and CO (now known as CO-w to distinguish it from the old, narrow CO-n filter) have been tested both on sky and with the dome flat lamps to establish the proper settings. Both the manual and the short list will be revised soon. For information, the lamp settings for 5s, Fowler1 are 1650 and 565 for the CN and CO-w filters, respectively, and 1080 and 375 for 16s, Fowler4.

IMPORTANT: The "Clear PAN" button on the shortcuts menu has been replaced by a "Stop PAN" button. When one starts up the PAN using the "Start PAN" button, it automatically carries out the cleanup functions which the "Clear PAN" button did, making that needlessly redundant. However, the "Stop PAN" button will clean up all of the background processes, so it should be used to close down the PAN at the end of the night, rather than simply closing the PAN window. Thanks to Behzad for implementing this feature.

23 January 2012: The new CN and CO filters were installed in WHIRC on 20 January, replacing the PaB4500 and old CO filters, respectively. The new CO filter will be given a designation (still TBD) to distinguish it from the old one. In the next couple of weeks we will be upgrading the MOP to display the new filter names and to write the new names to the image header. The filter scans will be added to the Filter webpage during that time as well.

9 January 2012: The new CN and CO filters arrived at the end of 2011. They have been inspected and scanned using the Lambda9 spectrophotometer. These results confirm the bandpass and transmission data from the vendor. The current plan is to install the new filters during the last two weeks of January.

6 January 2012: WTTM has become more problematic in the recent past, to the point where it should be considered a shared-risk enterprise. The primary issue appears to be that the X-Y stage lead screws are low-backlash high friction devices, and the stepper motors powering them are not sufficiently powerful to overcome the friction, particularly during cold weather. Charles Corson has experimented with various cleaning and lubrication schemes which appear to produce short-term improvement, but long term reliability will require a rebuild of the lead screws or the use of higher torque motors. Neither of these is likely to occur soon in the current resource environment.

Observers are cautioned against executing commands or motions to windows/devices which are in active use at the time. Adjusting the contrast/brightness of an image on the ds9 or attempting to open an image at the same time WHIRC is writing an image in autodisplay mode can crash the ds9 and result in a segmentation violation. Similarly, attempting to abort or pause an exposure while the array is in the process of reading out can crash the MOP and possibly require re-powering of the DHE, a process which we prefer to avoid.

Finally, it was noted that the Manual does not specify that the flatfield lamp settings are for the HIGH lamps. This will be added in the next revision of the manual.

26 October 2011:Recent experience with WTTM has shown that telescope offset motions under WTTM control often fail if the probe motion is outside of the nominal WHIRC field of view. Even though the mechanical travel of the probe stage should allow for a 25 arcsec buffer around the WHIRC field, the probe does not appear to offset reliably in this surrounding border area. The cause of this may be a mechanical issue with the XY stage near the limits of travel, but is currently unknown. We are strongly recommending that WTTM users ensure that the reference star stays within the WHIRC field of view to avoid this problem.

The current delivery schedule for the two WHIRC replacement filters is in mid-December, so the installation will probably not occur before late January 2012, and only then if the resources are available to carry out the filter change. Updates on the delivery and installation will be posted as they are known.

14 September 2011: As part of a science initiative by Matt Bershady, we will be replacing two of the WHIRC filters late this fall. The two replacement filters, which will be permanent contributions to the WHIRC filter complement, are designed to study the gravity-sensitive CO and CN bands in cool stars as a discriminant for determining luminosity classes in both galactic star-forming regions and in nearby galaxies. The two filters will be a somewhat broader and redder narrowband CO filter, which will replace the current CO filter, and a CN narrowband filter at 1.1 microns.

The new CO filter will be centered near 2.34 microns with a bandwidth of approximately 0.1 micron and should include the 2-0, 3-1, 4-2, and 5-3 CO first overtone bands. Although the thermal background will be higher than with the current CO filter (which was rarely used), the new filter will include more of the overall CO absorption. It is also closer to the narrowband CO filters which had been used in single-channel photometric observations of late-type stars. The new CN filter, in combination with the Low-airglow and HeI filters, can discriminate between carbon-rich and oxygen-rich AGB stars by virtue of their sampling the VO and TiO band heads in the latter.

Based on the usage of the current filters, we have decided that the CN filter will replace the redshifted Paschen beta filter. Since there are only two narrowband filters in the H band, we felt that both of the [FeII] filters should be retained. In the K band, the redshifted Brackett gamma filter serves as a continuum narrowband filter for both the Brackett gamma and H2 filters, and will also provide a continuum point for the new CO filter. While this could potentially impact some future science programs, we felt that replacing the least-used filter to be the best choice. Moreover, this is not necessarily a permanent change, as the redshifted Paschen beta (or another new filter required for a specific science program) may be installed at a future date.

17 February 2011: Dave Mills has added a feature to the MOP scripts which allows the user to pause the script at that point. This can be useful for acquiring guide stars after a large offset to or from a sky position, for example. Entering 'informUser [text message]' will pop up a dialog box with the [text message] and wait for the observer to acknowledge before proceeding with the script.

The latest version of the WHOMP (v1.82; 20 October 2010) was tested during the T&E run recently. The previously mentioned issues of incorrectly displaying negative declinations on the ds9 and the lack of correction for cos(dec) in calculating RA offsets have now been addressed. There may be some issues with the viewing of dither/map scripts on the ds9, depending on the computer operating system and ds9 version. The roll angle entry still results in rotation to the west of north (e.g., entering roll = 30 will yield the instruction 'rotatorAngle -30' in the script and the actual image will show the field rotated 30 degrees to the W of N).

08 December 2010: IMPORTANT!: The recent T&E testing of astrometric fields yielded WCS values which are rotated by approximately 0.8 degrees from those determined previously in 2008. The X and Y axes of the WHIRC array are now almost exactly (within 0.05 degrees) aligned with the cardinal celestial axes. A search of archival data suggests that this change occurred sometime between June and November 2009. As a result, I have updated the 'wprep.cl' task, which carries out various preparatory steps for data reduction such as Fowler renormalization, linearity correction, trimming the unused reference pixels, and correcting the WCS for any rotator offset. The new 'wprep.cl' task will now write 'WPREP = 2' into the header. The previous task is now labeled 'oldwprep.cl' and will write 'WPREP=1' to the header. No other changes have been made to the task. Since observers doing accurate astrometry will want to carry out a WCS mapping on their actual data, this change is mostly cosmetic, but it will give a much better initial alignment of the detector to the celestial axes. Data taken prior to June 2009 and not yet reduced should use 'oldwprep.cl' if one wants to get the axes close.

The reason for this change in the rotator offset zero point has not been determined.

12 November 2010: The Users Manual has been updated (again) to incorporate the new detector power/bias operation, as well as some diagnostic exercises from the PAN pull-down menu on the MOP. The new version is V1.10, 12 November 2010. The Summary Short List has also been updated to reflect the new startup procedures. We are planning on updating the WTTM GUI during the upcoming T&E run, so there will probably be another Manual revision in the near future.

26 October 2010: We have made some significant changes to the MOP observing interface. The detector power/bias operations are no longer carried out by the OA, but by the observer. When the MOP is started up, the Detector Biasing Window now automatically appears and the observer will carry out the Power Detector and Bias Detector steps. There are automatic safeguards which will not permit powering if any of the voltages are outside of a predetermined range or if the detector temperature is greater than 100 K. Moving this operation to the MOP is more efficient and in line with the operation of other instruments such as NEWFIRM; in addition, it eliminates the need for a socket connection which has been a source of problems in the past. The WTTM pull-down menu has been removed from the MOP, as it duplicates functions on the WTTM Client window. The User's Manual is being updated to reflect these changes.

Karen Butler carried out a careful test of the flatfield lamps and has determined lamp settings for the suggested signal levels for Fowler-1 (10000 ADU, 5 s) and Fowler-4 (40000 ADU, 16s) operation. These have been incorporated into the updated Summary List available on the WHIRC webpage.

29 September 2010: Now that WTTM appears to be working fairly reliably after the T&E in March and a couple of observing runs using it, we are encouraging observers to try using WTTM if it is consistent with the goals of their science programs. WTTM use will require a sufficiently bright (R &le 15) reference star within the WHIRC field of view and a planned dither/map routine which will maintain the reference star within the WTTM probe range, which is slightly larger than the WHIRC field. WTTM does appear to improve the image quality by about 0.1 arcsec in good seeing conditions and can markedly reduce the effects of wind shake on the telescope. Frequent usage will also help maintain staff proficiency.

There are still some issues with the WHOMP which users should keep in mind. Entering coordinates into the RA and DEC fields when defining a script can be useful in displaying the planned dither/map on a DSS field in the ds9 display. However, there is a bug/feature with the DEC input for negative declinations which requires a minus sign to be inserted in each of the three fields (degrees, minutes, seconds). Entering the minus sign only in the degree field (the conventional format) will result in the minutes and seconds being considered as positive numbers. In addition, while the dither map will be correctly displayed on the ds9, the RA offsets in the script output will be uncorrected for the cos(dec) factor and thus will not be correct (i.e., a 20 arcsec x 20 arcsec grid at a declination of 60 degrees will be output as a 40 x 20 arcsec grid). Until this is repaired, use RA, DEC of 00:00:00. for generating offset scripts.

13 April 2010: A new script, "decfix.cl" can be downloaded from the Data Reduction webpage. This script fixes the bug which caused an incorrect conversion from the sexigecimal "DEC" header value to the decimal "CRVAL2" value for negative declinations for data obtained prior to June 2009. The task must be sourced in IRAF with "task decfix = decfix.cl" and can operate on data in place or create new images with a desired extension. There is no effect on data with positive declinations or taken after June 2009.

07 April 2010: New versions of the User Manual and Short List are now downloadable from the webpage. The Short List has been modified to include the need to execute the Clear Pan command at the end of the night to ensure that the logging and other PAN daemons are shut down. The User Manual has been updated to reflect the new Mac-Mini operating platform, but should still be considered a DRAFT version until updated screen shots of the new window architecture have been obtained.

26 March 2010: The night of 24 March was spent carrying out more testing of WTTM. The primary issues identified from the November T&E night were the inability of the APD guider to hold onto the reference star, even when using the very gradual "guided dithers" mode. WTTM also appeared to successfully carry out some, but not all guide motions.

We made three significant improvements/discoveries during this run. 1. Charles carefully looked at the fiber coupling to the APDs and noted that even a slight axial shift of a fiber would seriously degrade the light coupling to the APD. He carefully worked the fibers into the mount and we now appear to have good signals. This does suggest that the previous problems with APD guiding had a strong hardware component. 2. The asymmetric plate scale of WHIRC on the WTTM optical system results in different pixel scales for the X and Y motions. This can result in missing the reacquisition of the guide star for long dither motions. The pixel scales in the WTTM software were adjusted to account for this. 3. We serendipitously discovered that some of the guided offsets were failing because the WTTM stage moved properly, but the telescope did not move at all. This was traced eventually to an error (feature?) in the parsing routine which accepts the offsetScope [x] [y] commands. If the inputs are fixed point or floating point with one or more significant figures (e.g., 10 or 10.0), then the offset works properly. However, a floating point with no significant figures (e.g., 10.) is not accepted and the telescope simply does not move and does not return an error message. After converting some of the ad hoc scripts which we had generated for testing WTTM (some of which had offsets in the offending format), to floating point with significant digits behind the decimal, we found that all of the dithering scripts worked properly. We exercised WTTM on a variety of scripts with amplitudes up to 75 arcsec with no failures. At this point, WTTM appears to be functioning properly.

01 March 2010: More updates on the Mac-Mini from Behzad: A few things about wiyn-1. It has two 24" monitors, a stereo speaker bar, webcam (ready for skype), and its own mouse and keyboard (i.e., totally independent of the old Linux setup with 3 monitors). The speaker bar is located under the right monitor, and it has a large knob on its right side to turn it on/off (when ON you see a blue light in the middle of the bar) and adjust the volume. The Mac Mini itself is sitting in between the two monitors for easy access to its DVD drive. Each monitor has two USB ports on its left side, to attach external disks or any other USB devices (note: no free USB port on the back of the Mac Mini). On the right hand side of the left monitor you will see a column of icons; some for an instrument (WHIRC, Hydra, MiniMo) and others for applications; the only two apps you should care about right now are "WIYN page" (if an observer is interested) and "Focus", which starts a tiny Tk widget showing the current Focus value (thanks to Di for the suggestion).

Wiyn-1 has 4 desktops (called spaces); if you move your mouse to the top right corner of the monitor on the right, or the bottom left corner of the monitor on the left, you will get a bird's eye view of all the 4 spaces and can pick one (or drag a window from one space to another). This is just a convenient feature, which some might actually find annoying, and can be easily changed (you can also switch to a different space by clicking on the "Spaces" icon in the Dock).

It will take some time to get used to a Mac and its nuances; one notable difference with Linux is that once you start an OSX application, terminating its open window does not kill the application; it just gets rid of the instance you were running (you will see the top menu bar still showing the app to be alive and well); and there are hot keys to start another instance, or kill the application. An odd ramification of this behavior is when you click on an icon in the Dock, expecting to start a new instance of that app in your current desktop, or space, like on Linux. but if that application is already running on another space, you will be automatically taken to that instance of the app! This is nice sometimes, and annoying other times :-) This default behavior can be modified a bit, but not much; bottom line: one has to learn some new tricks when using a Mac.

This is a work in progress, so there is plenty of room for improvement, and we still need to add a lot of features (e.g., in the long run we want to use wiyn-1 for remote observing, and also see if people can use the 24-bit ximtool for Mac to work with their images via NFS). Please provide feedback, and let me know if you have any questions or problems.

25 February 2010: As an experiment, we operated WHIRC using a Mac-Mini platform during last night's observing start. This configuration worked very well and will likely be the default in the future. The Mac-Mini provides two large high-resolution monitors, compared to the three smaller monitors used previously. The WIYN instruments (WHIRC, MiniMo, HYDRA) are launched by icons on the Mac-Mini screen, which initiates a VNC session to the appropriate computer. Thus it is no longer necessary to hardwire the monitor screens, mouse, and keyboard to the appropriate host computer. Once the VNC session to dust is opened, operation of WHIRC is unchanged from before. The User Manual will be updated to include screen shots of the new configuration.

The new Mac-Mini is named wiyn-1 (wiyn-1.kpno.noao.edu), which one can log into as 'observer', using the Observer password. Right now wiyn-1 mounts sand:/data1 and dust:/data1; you can access them in wiyn-1:~observer/data, where you see two directories, sand and dust. One use of this is to let observers run ds9 or the 24-bit ximtool on wiyn-1 (both already installed) and look at their data on sand or dust. Another is to let observers transfer their data more easily, either by burning a DVD using the Mac's drive or copying to their external USB disk, which they connect to one of the 4 USB ports on the two monitors.

Observers can always use scp to easily transfer any data file from any of our servers to any other server running Mac or Linux.

21 January 2010: As noted in the manual and on the website, the actual filters used in WHIRC are a nonstandard diameter and were therefore not measured by the vendor at cryogenic temperatures. Instead, the vendor measured a 25 mm diameter witness sample at both 300 K and 77 K and applied the measured shift for this sample to the 300 K scans of the actual WHIRC filters. This vendor information for all 13 WHIRC filters has been incorporated into a spreadsheet available from the Filters webpage. This is also a source of machine-readable data for those who wish to make their own transmission plots.

13 January 2010: An observer brought up a data reduction issue which will exist for all WHIRC (and possibly other WIYN instrumentation) data taken prior to June 2009. Because of a bug in the wcs.tcl module, fields at negative declinations will be displayed incorrectly on the ds9 (or other) displays. In the conversion from the (correct) sexigecimal DEC header value to the digital degrees CRVAL2 value, only the degrees are given the negative value. Thus, a DEC of -04:15:00 will be incorrectly rendered as CRVAL2 = -03.750000 (-03:45:00). Note this is the same issue which exists in the WHOMP, as noted in an earlier posting.

5 January 2010: Updated the WHIRC webpages.

  • Revised the User Manual (v 1.08) to include an expanded discussion of the WHOMP and a new section on Troubleshooting. Also updated the User Information webpage.
  • Finished a report on the mean-variance analysis of the WHIRC detector (WHIRC_VI_Mean-variance_100105.pdf) and posted on the Reports webpage.
  • Behzad has repaired the WHOMP to send the correct position angle to XTCS. The RA offsets for nonzero declination are still in error, and the ds9 viewer calculates negative declinations incorrectly.

    16 December 2009: Updated the WHIRC webpages.

  • Updated the WHIRC home page to include a link to this webpage.
  • Added a link on the WHIRC home page to the file "whomp.tar.gz", a version of the WHOMP observation planner which can be downloaded.
  • Modified the Checkout Manual and Summary List to reflect the replacement of the MASSIVE CLEANUP button with the clear PAN button.
    New versions of the User's Manual and User Information webpage are under construction to incorporate the revised detector gain and operation procedures.

    9 December 2009: Behzad converted all of the functions contained in the MASSIVE CLEANUP button to the clear PAN button, so there is now only one button on toolbar to carry out necessary recovery procedures. He has also removed the temperature monitoring feature on the Bias Tool as a step towards the eventual elimination of that tool.

    1 December 2009: A reanalysis of the Photon Transfer Curve data from September 2008 and additional data from the T&E night on 4 November 2009 resulted in a new value for the conversion gain. Analysis using the sum/difference method employed for the SNAP detectors instead of the traditional method of generating mean and sigma images using the IRAF 'imcombine' task appears to yield a much lower value for the gain (3.3 vs 3.8 - 4.0 e/ADU). Tests with artifically generated model data showed that the sum/difference method yielded the correct input gain, whereas the traditional method gave values 5 - 10% higher. Thus we will adopt the lower value of 3.3 e/ADU.

    5 November 2009: The previous night was dedicated to T&E of WHIRC and WTTM, with frustrating results. The startup, acquisition, and tip/tilt functions of WTTM appeared to work very well, on a variety of fields. However, offsetting for dither motions typically resulted in loss of the guide star after less than three motions, even if one was using the "guided dithers" where the APD probe slowly drags the telescope along while continuously guiding. This had worked well in the past. One possibility is a hardware problem with one or more of the APDs themselves. Charles will investigate this when he has time, but until then, WTTM operation has to be regarded as unreliable and is discouraged.

    The night was also marked by numerous DHE hangups. Post-run evaluation of the logs showed that most failures were a result of multiple observation commands. This could have resulted from a defective mouse which generated multiple click signals even if the button was pushed once. Dave Mills updated the MOP to ignore any commands once an integration has begun. The frequency of DHE hangups in succeeding observing nights has decreased drastically.



    rjoyce@noao.edu
    26 June 2014